Adverbs and Functional Heads: A Cross-Linguistic Perspective 9780195354058, 0195354052

Citation preview

Adverbs and Functional Heads

OXFORD STUDIES IN COMPARATIVE SYNTAX

Richard Kayne, General Editor

Principles and Parameters of Syntactic Saturation Gert Webelhuth Verb Movement and Expletive Subjects in the Germanic Languages Sten Vikner

Parameters and Functional Heads: Essays in Comparative Syntax Edited by Adriana Belletti and Luigi Rizzi Discourse Configurational Languages Edited by Katalin E. Kiss Clause Structure and Language Change Edited by Adrian Battye and Ian Roberts Dialect Variation and Parameter Setting: A Study of Belfast English and Standard English Alison Henry Parameters of Slavic Morphosyntax Steven Franks

Particles: On the Syntax of Verb-Particle, Triadic and Causative Constructions Marcel den Dikken The Polysynthesis Parameter Mark C. Baker The Role of Inflection in Scandinavian Syntax Anders Holmberg and Christer Platzack

Clause Structure and Word Order in Hebrew and Arabic: An Essay in Comparative Semitic Syntax Ur Shlonsky Negation and Clausal Structure: A Comparative Study of Romance Languages Raffaella Zanuttini Tense and Aspect: From Semantics to Morphosyntax Alessandra Giorgi and Fabio Pianesi

Coordination Janne Bondi Johannessen Adverbs and Functional Heads: A Cross-Linguistic Perspective Guglielmo Cinque

Adverbs and Functional Heads A Cross-Lingulstic

Perspective

GUGLIELMO CINQUE

New York Oxford OXFORD UNIVERSITY PRESS 1999

Oxford University Press Oxford New York

Athens Auckland Bangkok Bogoté) Bombay Buenos Aires Calcutta Cape Town Chennai Dares Salaam Delhi Florence Hong Kong Istanbul Karachi Kuala Lumpur Madrid Melbourne Mexico City Mumbai Nairobi Paris S&oPaulo Singapore Taipei Tokyo Toronto Warsaw and associated companies in

Berlin Ibadan

Copyright © 1999 by Guglielmo Cinque Published by Oxford University Press, Inc. 198 Madison Avenue, New York, New York 10016 Oxford is a registered trademark of Oxford University Press All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without the prior permission of Oxford University Press. Library of Congress Cataloging-in-Publication Data Cinque, Guglielmo. Adverbs and functional heads : a cross-linguistic perspective / Guglielmo Cinque.

p. cm. — (Oxford studies i comparative syntax) Includes bibliographical references and index. ISBN 0-19-511526-0; ISBN 0-19-511527-9 (pbk.)

1. Grammar, Comparative and general—Adverbials. 2. Order (Grammar) 3. Hierarchy (Linguistics) 4. Grammar, Comparative and

general—Clauses. [. Title. II. Series P284.C56 1998

415—dc21 98-21411

987654321 Printed in the United States of America on acid-free paper

Preface

This monograph has two interrelated goals (though their relation may not be immediately obvious). The first is to motivate an analysis of adverb phrases (AdvPs) as the unique specifiers of distinct maximal projections, rather than as adjuncts. The second is to argue for the existence of a fixed universal hierarchy of clausal functional projections. Despite the severe restrictions on phrase structure and movement proposed in Kayne (1994) and Chomsky (1995), U(niversal) Grammar) is often still assumed to allow wide variation among languages in the number and type of functional projections that they admit and/or in their relative order. Moreover, it is often assumed that in a single language, different clause types may instantiate different sets of functional projections. Here I try to construct a plausibility argument against these assumptions, suggesting that no such variation is allowed by UG and that the same number, type and order (hierarchy) of functional projections holds across languages and clause types, despite apparent counterevidence. Of course, to determine it empirically in detail is another matter, and what I have to say here is only a first approximation.

Specifically, | argue that in addition to the order of free functional morphemes (“particles” and auxiliaries) and of bound functional morphemes (affixes), there is a third important source of evidence for determining the hierarchy of functional projections—namely, the order and the nature of the different classes of AdvPs in the clause. We shall see that the different classes of AdvPs enter into a transparent Spec/ head relation with the different functional heads of the clause, providing evidence that may in certain languages be missing from the heads’ side and that, when present, converges with that deriving from the order of free and bound functional morphemes.

In other words, my suggestion is that adverbs are the overt manifestation of (the

vi PREFACE specifiers of) different functional projections, which in certain languages may also manifest themselves via overt material in the corresponding head positions. The first step in this plausibility argument is showing that the AdvPs of each class fill the unique Spec position of a distinct maximal projection. Crucial evidence for this conclusion is discussed in chapters 1 and 2, mainly on the basis of Romance data. In chapter 1, the fixed relative order of the different classes of AdvPs is established. In chapter 2, I argue that the distribution of past participles and finite verbs in Italian provides evidence for one head position to the immediate left and one head position to the immediate right of each AdvP in the fixed sequence. If sound, such an interpretation of the facts, in turn, constitutes strong evidence for locating each AdvpP in the unique Spec position of a distinct maximal projection (rather than in an adjunction position or in the Spec of a maximal projection hosting multiple Specs). The second step in the argument is establishing the hierarchy of the functional heads of the clause on independent grounds—namely, on the basis of the order of free and bound functional morphemes in different languages. This is attempted in chapter 3. Chapter 4 develops the third, and crucial, step in our plausibility argument, by matching the two independently established hierarchies and by showing the systematic one-to-one relation between the different AdvPs and the different functional heads. The other chapters are devoted to the discussion of certain extensions (such as the positions of AgrPs and NegPs, in chapter 5) and certain implications of the analysis (chapter 6). Chapter 7 briefly summarizes the main conclusions. This work began in 1992, prompted by the desire to better understand the functional projections hosting APs in the DP. The relative poverty of functional morphology on nouns offered little insight into the question, so the natural move was to see whether sentences provided a clearer picture of the projections hosting adverbs, the sentential counterpart of adjectives. The first results were presented in classes at the University of Venice in 1993 and at the Girona Summer School in Linguistics in 1994. Further elaborations were presented at the Glow conference

in Troms¢ in 1995 and at the Universities of Rome, Stuttgart, Bergamo, Paris, Vienna, McGill, and Amsterdam tn 1995 and 1996. I am indebted to those audiences and to many other people for comments, criticism, and references. I have tried to remember and thank them at the beginning of each chapter.

Venice G. C.

October 1997

Contents

Abbreviations XI

1 On the Relative Order of Adverb Phrases 3 1.1 “Lower” (pre-VP) AdvPs in Italian and French 4 1.2 “Higher” (sentence) AdvPs in Italian and French 11 1.3 “Lower” (pre-VP) AdvPs in VP-final position 13 1.4 Cases of AdvP movement and questions of scope 16 1.5 Circumstantial adverbials of place, time, manner, and the like 28 1.6 “Focusing” and “parenthetical” uses of AdvPs 30 1.7. Toward a universal hierarchy of AdvPs: some cross-linguistic evidence 32

2 A Case for Adverb Phrases in Spec Ad 2.1 Active past participle movement in Italian 45

2.2 Finite V movement in Italian 49 3 On the Order of Clausal Functional Heads 52 3.1 Evidence from the order of “nonclosing” (agglutinating) suffixes 53 3.2 Evidence from the order of “closing’”’ (inflectional) suffixes and auxiliaries 57

3.3. Evidence from the order of functional particles 58

3.4 Evidence from mixed cases 66 3.5 Some remarks on prefixes, derivation, and inflection 68 3.6 Toward a universal hierarchy of functional heads (a first approximation) 71 4 Matching and Refining the Hierarchies of Adverb Phrases and Functional Heads 77

4.1. Moods and modals 78 4.2 Theories of tense: evidence for Vikner’s (1985) three-relations theory 81

Vili CONTENTS

4.3 “Lexical” and “grammatical” aspect 83 4.4 Speech act adverbs and speech act mood 84 4.5 Evaluative adverbs and evaluative mood 84 4.6 Evidential adverbs and evidential mood 85 4.7 Epistemic adverbs and epistemic modals 86 4.8 Time adverbs and T(Past), T(Future) 87 4.9 “Perhaps” and irrealis mood 88 4.10 “(Not) necessarily/possibly” and alethic modals 89 4.11 Subject-oriented adverbs and root modals $9 4.12 Habitual adverbs and habitual aspect 90 4.13 Repetitive/frequentative adverbs and repetitive/frequentative aspects (I) 91 4.14 “Quickly/rapidly” and celerative aspect (1) 93

4.15 “Already” and T(anterior) 94 4.16 “No longer’ and terminative aspect 94 4.17 “Still” and continuative aspect 95 4.18 “Always” and perfect/imperfect aspect (7?) 96 4.19 “Just,” “soon,” and retrospective and proximative aspects 96 4.20 Durative adverbs and durative aspect 98

4.21 “?” and generic/progressive aspect 99 4.22 “Almost/imminently” and prospective aspect 99 4.23 Completely and tutto, and the two types of completive aspect 100 4.24 “Well” (manner adverbs) and voice 101 4.25 “Quickly/fast/early” and celerative aspect (ID 103 4.26 “Completely” and completive aspect (II) 104 4.27 Repetitive/frequentative adverbs and repetitive/frequentative aspect (ID) 104 4.28 Speculative remarks on other aspects and adverb classes 105 4.29 Toward a universal hierarchy of clausal functional projections

(a second approximation) 106 5 DP-Related Functional Projections and Negative Phrases 108

5.1 The positions of subject DPs 110 5.2 The positions of object DPs 115

5.3 Floating quantifiers 116 5.4 The positions of Neg(ative) P(hrase)s 120 6 Some Implications and Residual Questions 127 6.1 Default and marked values: simple and complex sentences 128 6.2. The hierarchy of functional projections and minimalist ideas 132 6.3. Semantics and the hierarchy of functional projections 134 6.4 Alleged parametric variation in the relative order of functional heads 136 6.5 Hierarchies of nonclausal functional projections 137

CONTENTS ix

7 Conclusions 140

Appendix 1 Some Remarks on Other Verbal Forms and Other Romance Varieties 142 A.1 __ Infinitives in French and Italian 143 A.2 Past participles in some Romance varieties 146 A.3 Absolute past participles, present participles, and gerunds 1n Italian 148 A.4_ Finite verbs in some Romance varieties 152 Appendix 2 A Synopsis of the Orders of Overt Functional Heads

in Individual Languages 153

Notes 167 References 231 Language Index 259

Name Index 263 Subject Index 267

This page intentionally left blank

Abbreviations

ABIL _ ability (modal, DEF definite suffix, ...) DETRANS | detransitivizer ABL _ ablative case DIR directional

ABS _ absolutive case DIST distantive (aspect,

ACC _ accusative case suffix)

AdvP Adverb Phrase DP Determiner Phrase

AGR agreement DUB dubitative (mood,

ANT _ anterior tense suffix)

AP Adjective Phrase DUR _ durative aspect

ASP aspect E event time Aux-to-COMP Auxiliary (raising) to EMPH — emphatic (particle,

COMP suffix)

CM Class marker EPISTEM ~ epistemic modality

COMITAT comitative ERG _ ergative COMP complementizer EVALUAT © evaluative (mood, COMPL completive aspect suffix, .. .)

CONAT _ conative aspect EVID _ evidential (particle,

COND _ conditional suffix, ...)

CONT continuative aspect F functional head

C(P) complementizer FEM feminine (Phrase) FQ floating quantifier DAT dative case FREQ _ frequentative

DEB ~— debitive (modal, aspect

suffix) FUT Future tense

DECL declarative (mood, GEN _ genitive case

suffix) HAB _ habitual aspect xi

xi ABBREVIATIONS

ILLOC _ illocutionary POSS _ possessive

(particle, suffix) POSSIB _ possibility (modal,

IMM = immediate aspect suffix) IMP imperative mood POT _ potential (modal, IMPERF imperfect aspect suffix)

INCEPT inceptive aspect PP Prepositional Phrase

INCH inchoative aspect P(REP) preposition INCONS _ inconsequential PRES present tense IND _ indicative mood PREV _ preverb INDEF _ indefinite PROBAB _probabilitative

INFL inflection (suffix, mood, .. .)

INGR _ ingressive aspect PROG progressive aspect INJ injunctive (mood, PROSP prospective aspect

particle) PROXIM _ proximative aspect INSTR instrumental case Q question (particle,

INT intensive suffix, ...) INTR intransitive QP Quantifier Phrase

IP Inflection Phrase QUOT = quotative (evidential)

IRR = irrealis (mood) suffix/particle

ITER _ iterative aspect R_ reference time LD _ locative/directional REFL _ reflexive

LF Logical Form REPET _ repetitive aspect LOC _ locative (preposition, RESULT _ resultative aspect

case,...) RETRO __ retrospective aspect

MASC masculine S speech time

MOD modal SEMEL — semelfactive

MOM = momentaneous SEQ _ sequential (suffix,

(aspect, suffix) aspect, ...)

NECESS __ necessitative SUBJ subjunctive

NEG _ negation ST — stative

NOM _ nominative case T(V/A) _ tense(/aspect)

NUM — number TAM _ tense/aspect/mood OPT optative mood TERMIN _ terminative aspect

PASS _ passive voice TOP _ topic

PAST _ past tense TR transitive

PERF _ perfect aspect UNR _ unrealized (aspect,

PERMISS _ permissive (modal, suffix)

suffix) VP Verb Phrase

PL plural

Adverbs and Functional Heads

This page intentionally left blank

1

On the Relative Order of Adverb Phrases

In this chapter, I try to establish the relative order of the main classes of AdvPs in Italian and French—an order that turns out to hold more generally in Romance languages (see chapter 2; Zanuttini, 1997, chap. 3; Paoli 1997, §3.1), and, from what we can gather from the limited evidence available (see §1.7), even cross-linguistically (a result of some interest, in itself, as itis not logically necessary). The argument that AdvPs enter into a fixed order (invariant across languages) requires explaining away those cases where they seemingly enter more than one order in one and the same language, or different orders in different languages. As we will see, typical sources of apparent counterexamples to the existence of a unique canonical order of AdvPs include the following:

1. When an AdvP directly modifies (is the specifier of) another AdvP. This may yield the opposite of the canonical order, but is clearly irrelevant, and can in general be told apart from the latter, as no material from the sentence can, in this case, intervene between the two adverbs. A number of such cases are discussed in §1.1. 2. When a lower portion of the clause (containing an AdvP) is raised across a higher AdvP (for focus-presupposition requirements). Cases of this sort are discussed in $1.3. 3. When one AdvP is wh-moved across another. Such cases are fairly obvious (and limited in application). They are discussed in §1.4. 3

4 ADVERBS AND FUNCTIONAL HEADS

4, When one and the same AdvP can be “base generated” in two different positions in the clause (with one of the two positions to the left, and the other

to the right of another AdvP). The different positioning of the AdvP can in general be detected from the different meaning or scope properties that the AdvP has in the two positions. Some such cases (which concern only certain classes of AdvPs) are discussed in §1.4 and in chapter 4. 5. When a noninherently “focusing” AdvP (e.g., probably) is used as a “focusing” adverb (like only or simply). In such usages, the AdvP can acquire different positions (and scopes) within the sentence. See $1.6. 6. When an AdvP is used “parenthetically” (see, again, $1.6). Evenifno real analysis will be proposed for such usages, they are intonationally quite clearly distinguishable from ordinary usages, and it should be relatively easy to keep them apart.

All six cases are argued here to be only apparent counterexamples to the existence of a unique, fixed, order of AdvPs. Although [ occasionally use the term adverb (for brevity), I do not assume it to be a head taking the VP, or some projection dominating the VP, as complement; in other words, I do not assume it to be part of the “extended projection” of V Gn Grimshaw’s 1991 sense). The evidence against this assumption ranges from the fact that adverbs do not block head movement of various verbal forms,’ to the fact that some of them can undergo Topicalization and Focus Movement, which are open to XPs but not to X°s. In the next chapter, I present an empirical argument for locating adverbs (more accurately, the Adverbial Phrases they head) in distinct specifier positions.?

1.1. “Lower” (pre-VP) AdvPs in Italian and French I begin by considering those AdvPs that occur in Italian in the lower portion of the clause, in the “space” delimited on the left by the leftmost position that an (active) past participle can come to occupy and on the right by a complement (or the subject) of the past participle. In French, the same space is not delimited on the left by the past participle, as this necessarily remains closer to its complements (.e., “lower’’) than in Italian. This difference, however, has no consequences for this comparison between the two languages. The relative order of the AdvPs occurring in this “space” appears to be rigidly fixed, as can be seen by considering the relative order of any two pairs of them. Habitual adverbs like solitamente ‘usually’ precede the negative adverb mica:* (1) a. Alle due, Gianni non ha solitamente mica mangiato, ancora. ‘At two, G. has usually not eaten yet.’ b. *Alle due, Gianni non ha mica solitamente mangiato, ancora. ‘At two, G. has not usually eaten yet.’

The same appears to be true in French. The habitual adverb généralement precedes pas and cannot follow it:

ON THE RELATIVE ORDER OF ADVERB PHRASES 5

(2) a. A deux heures, Gianni n’a généralement pas mangé, encore. b. *A deux heures, Gianni n’a pas généralement mangé, encore.”

In Italian, the negative adverb mica necessarily precedes the adverb gid ‘already’ :® (3) a. Non hanno mica gia chiamato, che io sappia. ‘They have not already telephoned, that I know.’ b. *Non hanno gia mica chiamato, che 10 sappia. “They have already not telephoned, that I know.’

The corresponding French adverbs pas and déja show an analogous relative order:’ (4) a. Situ n’as pas déja mangé, tu peux le prendre. ‘If you have not already eaten, you can take it.’ b. *Si tu n’as déja pas mangé, tu peux le prendre. ‘If you have already not eaten, you can take it.’

The adverb gia necessarily precedes the adverb piu ‘any longer’: (5) a. Al?epoca non possedeva gia piu nulla. ‘At the time (s)he did not possess already any longer anything.’ b. *All’epoca non possedeva piu gia nulla. ‘At the time (s)he did not possess any longer already anything.’

The same holds for French (see Togeby 1984, 259): (6) a. Al’époque, il ne possédait déjd plus rien. b. *A l’époque, il ne possédait plus déja rien.

Given that mica precedes gia and gia precedes piu, by transitivity we correctly expect mica to precede pitt: (7) a. Non hanno chiamato mica piu, da allora.® ‘They haven’t telephoned not any longer, since then.’ b. *Non hanno chiamato pi mica, da allora. ‘They haven’t telephoned any longer not, since then.’

The analogous expectation that pas in French precedes plus is at first sight not borne out. The two cannot co-occur (in either order): (8) a. *Ils n’ont pas plus téléphoné. ‘They haven’t not any longer telephoned.’ b. *Ils n’ont plus pas téléphoné. ‘They haven’t any longer not telephoned.’

We might think that they cannot co-occur because they occupy the same position in the clause. But this is not correct. First, the fact that pas precedes déja, and plus follows déja, argues against identifying the two positions. Such relative orders show that pas occupies a position higher than plus. A second piece of evidence for taking pas to be higher than plus comes from the syntax of infinitives. As Pollock (1989, 413) notes, for some speakers, a lexical infinitive can precede plus (as well as follow

it): Ne dormir plus ... (and Ne plus dormir ...) ‘Not to sleep any longer...’ But apparently no speaker allows a lexical infinitive to precede pas: *Ne dormir pas... versus Ne pas dormir ‘Not to sleep .. .”?

6 ADVERBS AND FUNCTIONAL HEADS

In the spirit of his verb movement analysis of infinitives in French, the systematic contrast just noted can be explained if pas is indeed higher than plus and the infinitive is able to raise to a head to the left of plus but is unable to raise any higher:!°

(9) [__[ pas_ [plus... dormir ] ] ] If they occupied the same structural position, their different behavior with respect to lexical infinitives would remain unexplained. So far, then, we have evidence for the following relative orders: (10) a. solitamente > mica > gia > pit!!! b. généralement > pas > déja > plus

Consider now the relative position of such adverbs with respect to sempre/toujours ‘always’. As the following sentences show, in Italian, sempre follows piu and, a fortiorl, gid, mica, and solitamente: (11) a. Daallora, non ha pit sempre vinto. b. *Da allora, non ha sempre piu vinto. ‘Since then, he has no longer always won.’

(12) Quando si presenta un problema... “When a problem arises .. .’ a. lui sa gid sempre come fare. ‘he knows already always how to act.’ b. *lui sa sempre gid come fare. ‘he knows always already how to act.’ !? (13) a. Gianni non ha mica sempre vinto. b. *Gianni non ha sempre mica vinto. ‘G. hasn’t always won.’

(14) a. Ha solitamente sempre ragione lui. ‘He is usually always right.’ b. *Ha sempre solitamente ragione lui. ‘He is always usually right.’ The same is true when three (or more) adverbs occur together: (15) a. Daallora, non accetta mica pi sempre i nostri inviti. ‘Since then, he doesn’t any longer always accept our invitations.’ b. *Da allora, non accetta mica sempre piu i nostri invit. c. *Da allora, non accetta sempre mica pid i nostri inviti. d. *Da allora, non accetta sempre pitt mica i nostri invitt. e. *Da allora, non accetta pia mica sempre i nostri inviti. f. *Da allora, non accetta pi sempre mica i nostri inviti.

Analogously, in French, toujours ‘always’ follows plus and cannot precede it: (16) a. A partir de ce moment 1a, il n’a plus toujours vaincu. b. *A partir de ce moment 1a, il n’a toujours plus vaincu. ‘Since then, he has no longer always won.’

By transitivity, as in Italian, toujours is also correctly predicted to follow déja, pas, and généralement:

(17) Quand il y aun probleme... ‘When there is a problem .. .’ a. il sait déja toujours comment faire. ‘he already always knows how to act.’ b. *il sait toujours déjia comment faire. ‘he always already knows how to act.’!?

ON THE RELATIVE ORDER OF ADVERB PHRASES 7

(18) a. Elle ne chantera pas toujours. ‘She will sing not forever.’ ! b. *Elle ne chantera toujours pas. ‘She will sing forever not.’ (19) a. C’est lui qui a généralement toujours raison. ‘It’s him who is usually always night.’ b. *C’est lui qui a toujours généralement raison. ‘It’s him who is always usually right.’

Sempre/toujours ‘always’ appears to necessarily precede, if present, an adverb like completamente/complétement ‘completely’: (20) a. Gianni ha sempre completamente perso la testa per lei. ‘G. has always completely lost his mind for her.’ b. *Gianni ha completamente sempre perso la testa per lei. ‘G. has completely always lost his mind for her.’ (21) a. Jean a toujours complétement perdu la téte pour elle. b. *Jean a complétement toujours perdu la téte pour elle.

The sequence in (22) gives the fixed relative order of the elements so far examined: (22) a. solitamente > mica > gia > pil > sempre > completamente b. généralement > pas > déja > plus > toujours > complétement

Consider now the relative order of completamente/complétement ‘completely’ , (unstressed) tutto/tout, and (unstressed) manner adverbs like bene/bien, male/mal, and SO on.

(Unstressed) tutto/tout precede (unstressed) bene/bien/male/mal, and so on, at least in the position we are focusing on here, which is the position preceding the complements of the participle (and the participle itself in French). See Kayne (1975, 26ff), from which (24) is drawn, and Sportiche (1988, 433): (23) a. Ha gia detto tutto bene Gianni. ‘Has already said everything well G.’ b. *Ha gia detto bene tutto Gianni. (irrelevantly possible with nuclear, or contrastive, stress on tutto, and with Gianni “de-accented.”’)

(24) a. Elle a tout trés mal compris. ‘She understood everything very poorly.’ b. *Elle a trés mal tout compris. ‘She understood very poorly everything.’

This, in fact, is the unmarked position of tutto and bene (and manner adverbs in general)—-see Lepschy and Lepschy (1977, 184)—unless they are modified, coordinated, or focused (see Kayne 1975, § 1.6; Lonzi 1991, 358ff; Cardinaletti and Starke 1994). If tutto is modified, coordinated, or focused, it can appear after bene: (25) a. Hanno spiegato bene pressoché tutto alla maestra. ‘They explained well almost everything to the teacher.’ b. Hanno spiegato bene tutto o quasi (tutto) alla maestra. ‘They explained well everything or nearly everything to the teacher.’ c. Hanno spiegato bene TUTTO, alla maestra. ‘They explained well EVERYTHING (focus) to the teacher.’

If bene is likewise modified, coordinated, or focused, it can occur after the complements of the participle:

8 ADVERBS AND FUNCTIONAL HEADS

(26) a. Hanno detto tutto alla maestra veramente bene. ‘They have said everything to the teacher really well.’ b. Hanno detto tutto alla maestra bene o quasi (bene). ‘They have said everything to the teacher well or almost well.’ c. Hanno detto tutto alla maestra BENE. ‘They have said everything to the teacher WELL.’

The former and latter positions of tutto/tout and bene/bien are explicitly equated in Cardinaletti and Starke (1994) to the “derived” and “base” positions of pronouns, respectively, which display an analogous pattern: (27) a. Janice called up the man/*him. b. Janice called the man/him up.

(28) a. Janice called up only him. b. Janice called up him and her. c. Janice called up HIM. (see Baker 1989, 156)

In each of these cases, only the strong variant (the one modified, coordinated, or focused) can apparently occupy the “base” position, while the other (the “weak” one) must occupy a special derived position. !® As to the relative order of completamente (parzialmente) / complétement (partiellement) ‘completely/partially’, tutto/tout ‘everything’, and bene/bien ‘well’, we observe that the first must precede the latter two: (29) a. Ha rifatto parzialmente tutto bene Gianni. ‘Has done again partially everything well G.’ b. *Ha rifatto tutto parzialmente bene Gianni.'’ ‘Has done again everything partially well G.’

(30) a. Uacomplétement tout perdu. ‘He lost completely everything.’ b. *?Il a tout complétement perdu."®

The overall order of the adverbs seen so far is shown in (31): (31) a. solitamente > mica > gia > pil > sempre > completamente > tutto > bene b. généralement > pas > déja > plus > toujours > complétement > tout > bien!”

Each adverb in (31) is representative of a larger class of adverbs, whose exhaustive list is beside the point here. I will limit myself to indicating few other members for each class (as remarked in the literature, no more than one member of each class can appear in a clause; Steinitz 1969, 5Off; Jackendoff 1972, 87; Quirk et al. 1985, 487ff). Other classes of “lower” adverbs not included here are discussed in chapter 4. In the same class of solitamente/généralement are adverbs like di solito, abitualmente, usualmente, and normalmente and habituellement, normalement, d’habitude, and ordinairement. Other negative adverbs that seem to occupy the same position as mica are affatto ‘(not) at all’, no ‘(emphatic) not’, neanche/nemmeno/neppure ‘not even’ .?° To the same class of gia ‘already’ belong poi ‘(literally) after’ andnon... ancora ‘not... yet’. Pot, like gia, follows mica and precedes piu (Non ha mica poi piu detto se veniva ‘He has not after any longer said if he was coming’).7! As expected, poi

ON THE RELATIVE ORDER OF ADVERB PHRASES 9

cannot co-occur with gid, in any order: *Non ha mica gia poi piu detto se veniva; *Non ha mica poi gia pi detto se veniva, Non... ancora ‘not yet’ is the negative counterpart of gid (see Pecoraro and Pisacane 1984, 54, and, for the English equivalents, Traugott and Waterhouse 1969).”7 It follows mica (Non V’ho mica ancora letto ‘T have not yet read it’ versus *Non l’ho ancora mica letto) and must, if anything, precede pitt (?Non ha ancora piu ricevuto nulla ‘He hasn’t yet any longer received anything’ versus *Non ha pit ancora ricevuto nulla ‘He hasn’t any longer yet received anything’ ).*° Another adverb belonging to the same class (and position) of pit 1s its positive

counterpart ancora ‘still’ (see Pecoraro and Pisacane 1984, 55, and Vikner 1978, 93ff, for French plus and encore).** Indeed, just like pia, it is found to the left of sempre: Lui ha ancora sempre il coltello dalla parte del manico ‘He still always has an advantage’ versus *Lui ha sempre ancora il coltello dalla parte del manico ‘He always still has an advantage.’ *®

A plausible candidate for the same class as sempre is its negative counterpart mai ‘(njever’, though at first sight, this appears unlikely, since sempre follows piu, as noted, whereas mai appears to have to precede it. Compare (32) with (33): (32) a. Lui non ha pit sempre vinto, da allora. ‘He has not any longer always won, since then.’ b. *Lui non ha sempre piu vinto, da allora. ‘He has not always any longer won, since then.’

(33) a. *Lui non ha pit mai vinto, da allora. ‘He has not any longer ever won, since then.’ b. Lui non ha mai pit vinto, da allora. ‘He has not ever any longer won, since then.’

However, we have seen evidence that the sequence mai piu is necessarily a constituent, with mai in the Spec of pia ([mai [pit]]), correctly predicting that no head posi-

tion is available for the past participle between mai and pit (see the discussion of (111) in note 16). This, plus the assumption that the sequence pi mai is, for some reason, blocked in modern standard Italian, makes it still possible to maintain that mai occupies the same position as sempre.?! In this respect, French more directly fulfills our expectations in that the unmarked sequence is indeed plus preceding jamais (just as plus precedes toujours), although the other order is also possible (albeit less common):78

(34) a. I1n’a plus jamais rien su d’elle. ‘He hasn't any longer ever learned anything about her.’ b. I n’a jamais plus rien su d’elle. ‘He hasn’t ever any longer learned anything about her.’

But, again, there is reason to believe that the two sequences are structurally rather different, the second having jamais necessarily in the Spec of plus, just as with the sequence mai piu of Italian. Evidence for this comes from certain observations made in Engver (1972). He reports (p. 24) that the infinitive may be found following plus jamais (the preferred option), or preceding it, or between plus and jamais. However, he reports no case in which the infinitive intervenes between jamais and plus. We

10 ADVERBS AND FUNCTIONAL HEADS

can take this to be a consequence of the fact that the sequence jamais plus, in contrast to plus jamais, has jamais necessarily tn the Spec of plus, which leaves no room for the infinitive between the two adverbs (the same reason that excluded the past participle between mai and piu in Italian).”? Just as gia precedes sempre, non... ancora, the negative counterpart of gid, is expected to precede mai, the negative counterpart of sempre. This is indeed what we find:

(35) a. Non te l’avevo ancora mai detto? ‘Hadn’t I yet ever told you?’ (N.Ginzburg, Ti ho sposato per allegria, Torino, Einaudi, 1966, 67).

b. *Non te l’avevo mai ancora detto. ‘T hadn’t ever yet told you.’ The same contrast is found in French, according to my informants (see (36a—b)), even though, in some contexts, the opposite order Gamais encore) 1s apparently also pos-

sible (see Je trouverai bien un endroit ou personne ne sera jamais encore venu ‘T will find a place where nobody will have never yet been’, cited in Togeby, 1984, 219):

(36) a. Je n’ai encore jamais lu ce livre. ‘[ haven't yet ever read this book.’ b. ??Je n’ai jamais encore lu ce livre. ‘[ haven’t ever yet read this book.’

The two sequences are expected to differ in structure, however. The one with the “unexpected” order Gamais encore) should have jamais in the Spec of encore. And the facts conform to the expectation. While an infinitive can intervene between encore and jamais (37a), no infinitive (nor any other material, for that matter) can intervene between jamais and encore (37b): (37) a. (?)N’encore étre jamais venu ici est inadmissible. ‘Not yet to have ever come here is not to be admitted.’ b. *Ne jamais étre encore venu ici est inadmissible. ‘Not ever to have yet come here ts not to be admitted.’

Other AdvPs filling the position of completamente are interamente, parzialmente, del tutto, in parte, and so forth, and their counterparts in French. Whether niente (and nulla) ‘nothing’ can fill the same position as tutto (just as rien in French is taken to fill the same position as tout; Kayne 1975, §1.3) 1s, at first sight, doubtful. Unlike tutto, niente can precede the light manner adverb bene only if it receives the most prominent stress (and bene, and whatever follows it, is de-accented):

(38) a. Ha fatto tutto bene Gianni. ‘Has done everything well G.’ b. *Non ha fatto niente bene Gianni. ‘Not has done anything well G.’ c. Non ha fatto NIENTE, bene, Gianni.

This, however, is not particularly revealing, given that the same effect 1s found when VP-final complements follow niente:

(39) a. *Non mandero niente a casa a Gianni. ‘I will send nothing home to G.’ b. Non manderé NIENTE, a casa, a Gianni.

This effect is suspended 1f another negative constituent is found in sentence-final position and receives the most prominent stress of the sentence:

ON THE RELATIVE ORDER OF ADVERB PHRASES 11

(40) Non mandero niente a casa a nessuno. ‘T will not send anything home to anybody.’

In exactly the same circumstances, niente (but no other negative phrase) can indeed precede the light manner adverb bene:*° (41) a. Non spiegd mai niente bene a nessuno. ‘He never explained anything well to anybody.’ b. *Non spiegd mai nessuna istruzione bene a nessuno. ‘He never explained any instruction well to anybody.’

The position occupied by bene (bien in French) in (41) appears to be a position for manner adverbs, and possibly a few other classes, like the measure adverbs molto/ beaucoup ‘much’, poco/peu ‘little’, and so forth:*! (42) a. Ha apprezzato tutto molto anche Gianni. ‘Has appreciated everything much G. too.’ b. *Ha apprezzato molto tutto anche Gianni. ‘Has appreciated much everything G. too.’

(43) a. Ha tout beaucoup apprecié. ‘He has everything much appreciated.’ b. *Ii a beaucoup tout apprecié. ‘He has much everything appreciated.’

We thus have the following classes of AdvPs coming in the relative order shown in (44): (44) a. solitamente> mica> gid> pil> sempre > completamente > tutto> bene*”

di solito neanche poi ancora mai parzialmente niente male abitualmente neppure non ancora

b. généralement pas déja plus toujours complétement tout bien habituellement (pas) encore encorejamais _ partiellement rien mal 1.2 “Higher” (Sentence) AdvPs in Italian and French Except for a well-defined apparent exception having to do with “speech—time” ad-

verbs, to which we shall return, a fixed relative order also characterizes higher adverbs. For example, according to Jackendoff (1972, 89), “subject-oriented” adverbs like intelligently and clumsily follow “speaker-oriented” adverbs like probably (see also Sueur 1978, 247). Jackendoff’s class of “speaker-oriented” adverbs is not homogeneous though, but conflates at least the following distinct classes, as shown on syntactic and semantic grounds by Bellert (1977): (45) a. domain adverbs: politically, legally* b. pragmatic adverbs: frankly, sincerely, honestly c. evaluative adverbs: luckily, fortunately, happily d. modal adverbs: probably, presumably*

€. perhaps

Although Bellert (1977) does not consider it additional evidence for subdividing, as she does, Jackendoff’s class of “speaker-oriented”’ adverbs comes from the obser-

vation that the adverbs of each category of (45) can indeed co-occur (in a certain order), which would be unexpected if they were members of the same class (by Jackendoff’s own criteria; see Jackendoff 1972, 87ff).

12 ADVERBS AND FUNCTIONAL HEADS

What we find is that besides preceding (as expected) “subject-oriented” AdvPs (see (46a-b)), forse ‘perhaps’ can follow “modal” adverbs like probabilmente (see (47a—b)):*>

(46) a. Gianni accettera forse saggiamente il vostro aiuto.*® ‘G. will perhaps wisely accept your help.’ b. *Gianni accettera saggiamente forse il vostro aiuto. ‘G. will wisely perhaps accept your help.’

(47) a. Gianni sara probabilmente forse ancora in grado di aiutarci. ‘G. will probably perhaps still be able to help us.’ b. *Gianni sara forse probabilmente ancora in grado di aiutarci. ‘G. will perhaps probably still be able to help us.’

“Modal” adverbs, 1n turn, have to follow “evaluative” adverbs like (s)/ortunatamente, per (s)fortuna ‘(un)luckily, and purtroppo ‘unfortunately’ :°7 (48) a. Gianni ha per fortuna probabilmente accettato. ‘G. has luckily probably accepted.’ b. *Gianni ha probabilmente per fortuna accettato. ‘G. has probably luckily accepted.’

This is noted, for French, in Sueur (1978, 238); for Dutch, in Koster (1978, 205ff), and, for German, in Doherty (1985, 112ff) (see also Bartsch 1976, 235ff), where contrasts like the following are given: (49) a. Heureusement, sans doute que Pierre viendra. ‘Luckily, undoubtedly P. will come.’ b. *Sans doute, heureusement que Pierre viendra. ‘Undoubtedly, luckily P. will come.’

(50) a. Het is zo dat hij helaas waarschijnlijk ziek 1s. ‘It is the case that he unfortunately probably sick is.’ b. *Het is zo dat hij waarschijnlijk helaas ziek 1s. ‘It is the case that he probably unfortunately sick is.’

(51) a. Konrad ist leider vermutlich verreist. ‘K. has unfortunately presumably left.’ b. *Konrad ist vermutlich leider verreist. ‘K. has presumably unfortunately left.’

As Koster and Doherty note, the relative order of the two AdvPs cannot even be altered by movement of one of the two to COMP; a question to which I return in § 1.4.8

“Evaluative” adverbs, in turn, follow “pragmatic” adverbs like francamente ‘frankly’ and sinceramente ‘sincerely’, which are also called “illocutionary” adverbs (see Vendler 1984) or “speech act” adverbs (see Roberts 1985a):°? (52) a. Francamente ho purtroppo una pessima opinione di voi. ‘Frankly I have unfortunately a very bad opinion of you.’ b. *Purtroppo ho francamente una pessima opinione di voi. ‘Unfortunately I have frankly a very bad opinion of you.’

Temporal adverbs anchored to speech time, like ora, adesso ‘now’ and allora ‘then’ seem to enjoy a partially freer distribution. Although they have to precede “subjectoriented” adverbs (see (53a—b)), and (preferably) forse (see (54a—b)), they can ap-

ON THE RELATIVE ORDER OF ADVERB PHRASES 13

parently either follow or precede “modal,” “evaluative,” and “pragmatic” adverbs. See (55)~(57a—b):

(53) a. Gianni ha ora saggiamente ceduto. ‘G. has now wisely surrendered.’ b. *Gianni ha saggiamente ora ceduto. ‘G. has wisely now surrendered.’

(54) a. Gianni é ora forse partito. ‘G. has now perhaps left.’ b. *?Gianni é forse ora partito. ‘G. has perhaps now left.’ (53) a. Probabilmente ora ci ascoltera. ‘(Syhe probably now will listen to us.’ b. Ora probabilmente ci ascoltera. ‘(S)he now probably will listen to us.’ (56) a. Fortunatamente ora sei con noi. ‘Luckily now you are with us.’ b. Ora fortunatamente sei con noi. ‘Now luckily you are with us.’ (57) a. Francamente ora mi hai stufato. ‘Frankly now you have annoyed me.’ b. Ora francamente mi hai stufato. ‘Now frankly you have annoyed me.’ This distribution makes sense if speech—time adverbs are generated to the left of forse (and “subject-oriented” adverbs) and to the right of “modal” (and “evaluative” and “pragmatic”) adverbs, but crucially, they can also appear in a (“Topic”) position to

the left of all sentence adverbs—plausibly in the position of “domain adverbs (see note 41). If so, speech—time adverbs appear either to the right or to the left of “modal,” “evaluative,” and “pragmatic” adverbs but can appear only to the left of forse and “subject-oriented” adverbs, in agreement with the observed facts.*! This gives us the following relative order for the “higher” adverb classes (I omit the preceding “Topic” position hosting “adverbs of setting”): (58) a. francamente > fortunatamente > evidentemente > probabilmente > ora >

sinceramente purtroppo chiaramente presumibilmente allora forse > intelligentemente per caso goffamente b. franchement > heureusement > évidemment > probablement > maintenant > peutétre > intelligentement

The sequence in (58) linearly precedes the sequence of “lower” adverbs in (44).

1.3. “Lower” (pre-VP) AdvPs in VP-final position In this section, I briefly consider the special option for some of the AdvPs in pre-VP position, of being found in VP-final position following the complement(s) of the verb. Because this is largely uncharted territory, my conclusions are necessarily very tentative. However, this discussion has at least a methodological virtue in that it exposes an important source of apparent exceptions and counterexamples to the hierarchy of AdvPs just proposed. Refer to (15a-f), which show the only permitted relative order of the AdvPs mica, piu, and sempre in pre-VP position. The impossible order sempre mica pitt of (15c) becomes unexpectedly possible if the object is cliticized and mica pit is more heavily stressed and separated from sempre by a slight pause: (59) Da allora, non li accetta sempre (#) mica PIU. ‘Since then, he doesn’t accept them always not any longer.’

14 ADVERBS AND FUNCTIONAL HEADS

I interpret this as an indication that mica pitt in (59) does not belong to the same preVP adverbial “space” to which sempre belongs, but to a distinct adverbial “space,” a VP-final one, following the complements (not shown in (59)). This is confirmed by comparing the impossible (15c), where the three AdvPs precede the complement,

with the possible (60), where sempre precedes the complement and mica and pit follow it (bearing some stress): (60) Da allora, non accetta sempre i nostri inviti mica PIU. ‘Since then, he doesn’t accept always our invitations not any longer.’

Now we can see better why (59) is also possible; because it is compatible with a structure analogous to (60), where the trace of the clitic is between sempre and mica piit. Virtually all the “lower’’ AdvPs that can appear in the “space” preceding the complements in Italian can also occur in the “space” following the complements (see (61)). Because they must be able to bear heavy stress when they occur 1n that “space,” they are most natural there when reinforced by a specifier, as is the case with the PIU modified by mica in (59) and (60). Some have even a morphologically heavier variant, which can

appear in precomplement position and is the one preferentially chosen in the postcomplement “space” (di gid, neancora—the latter possible only in northern varieties): (61) a. *Gianni non ha vinto la lotteria MICA (0 quasi mica). ‘G. has not won the lottery not (or almost not).’ b. Gianni ha ricevuto la notizia *(DI) GIA. ‘G. has received the news already.’ c. Gianni non legge l’alfabeto (NEYANCORA.

‘G. does not read the alphabet yet.’ a d. Gianni non vince le partite mica PIU/PIU o quasi PIU. ‘G. does not win his matches not any longer/any longer or almost any longer.’

e. Gianni vede Maria ANCORA. ‘G. is seeing M. still.’ f. Gianni ha dato a Maria TUTTO. ‘G. has given to M. everything.’ g. Gianni ha rifatto i compiti BENE. ‘G. has redone his homework well.’

When they occur in the postcomplement “space,” the “lower” AdvPs of $1.1 appear to obey the same ordering restrictions operative in the pre-VP “space.” See, for example, (62): (62) a. Gianni non vince le sue partite gid pid sempre BENE. ‘G. does not win his matches already any longer always well.’ b. *Gianni non vince le sue partite gid sempre piu BENE. c. *Gianni non vince le sue partite gia bene piu SEMPRE. d. *Gianni non vince le sue partite sempre gia piu BENE. e. *Gianni non vince le sue partite gia piu bene SEMPRE.

It seems, then, that the only “exceptions” to this rigid order of AdvPs arise when we mix AdvPs from both “spaces.” But clearly, this does not compromise the existence of the rigid order noted (now relativized to each “space’’). In the next section, we take up the question of how best to relate these two “spaces.” If we extend our examination to the “higher” (sentence) AdvPs of §1.2, we find that they (and habitual adverbs) cannot occur in the postcomplement “space” unless they are “de-accented.”’ Compare (63) and (64):

ON THE RELATIVE ORDER OF ADVERB PHRASES 15

(63) a. *Non posso sopportare neanche Carlo onestamente. ‘I can’t stand C. honestly.’ b. *Mario si é rimesso dalla sua malattia fortunatamente. ‘M. recovered from his iliness luckily.’

c. *Prendero il treno probabilmente. ‘I will get the train probably.’ d. *Vedro Gianni forse. ‘I will see G. perhaps.’ e. *Ho aspettato Gianni saggiamente. ‘I waited for G. wisely.’ f. *Gianni beve vino solitamente. ‘G. drinks wine usually.’ (64) a. Non posso sopportare neanche Carlo, onestamente. ‘I can’t stand C., honestly.’ b. Mario si é rimesso dalla sua malattia, fortunatamente. ‘M. recovered from his illness, luckily.’

c. Prender6 il treno, probabilmente. ‘I will get the train, probably.’ d. Vedr6é Gianni, forse. ‘T will see G., perhaps.’ e. Ho aspettato Gianni, saggiamente. ‘{ waited for G., wisely.’ f. Gianni beve vino, solitamente. ‘G. drinks wine, usually.

In fact, more than one de-accented sentence AdvP can appear after the constituent bearing the nuclear stress of the sentence (or a focus stress). In this case, no rigid relative order 1s found: (65) a. Sembra che lascera anche noi/NOI, purtroppo, forse. ‘It seems that he will leave us too, unfortunately, perhaps.’ b. Sembra che lascera anche noi/NOI, forse, purtroppo. (66) a. Non sopporto NESSUNO, francamente, di solito. ‘I can’t stand anybody, frankly, usually.’ b. Non sopporto NESSUNO, di solito, francamente.

We return to a possible implication of these facts in the next section. An exception is again provided by “speech—time” adverbs (which we earlier saw occurring among the “higher” (sentence) AdvPs), for these can also occur in postcomplement position without being de-accented: (67) Hanno dato la notizia a Gianni proprio ora / allora / and so forth. ‘They gave the news to G. just now / then / and so forth.’

Just as they were seen to occur also in the “adverb of setting” position (as instances of the larger class of temporal adverbials), I take them to also occur, for the same reason, in the other typical position of temporal adverbials, the one after the complements in VP. This postcomplement position of temporal adverbs is just one among

many other classes of adverbs and adverbial PPs and also embedding clauses (adverbials of place, time, manner, means, reason, purpose, and so on). These postcomplement VP-internal adverbials—as opposed to the classes of AdvPs discussed so far—do not seem to be rigidly ordered (see Chomsky 1995, 333). As shown in (68), all arrangements seem to be possible, except for, perhaps, differences in scope—a significant property that sets them apart from the classes of AdvPs considered so far. (I return to this class of elements in §1.5.) (68) a. Seguiro le lezioni tutti i giorni all’universita diligentemente. ‘{ will attend classes every day at the university with great zeal.’

16 ADVERBS AND FUNCTIONAL HEADS

b. Seguiro le lezioni all’universita tutti i giorni diligentemente. c. Seguiro le lezioni all’universita diligentemente tutti 1 giorni. d. Seguiré Ie lezioni diligentemente all’ universita tutti i giorni.

As (69) shows, when the lower pre-VP AdvPs appear after the complements, they also follow all the (unordered) temporal, locative, manner, and so on adverbials, unless

the latter are de-accented:* (69) a. Non seguird le lezioni tutti i giorni all’ universita diligentemente mai PIU. ‘T will attend classes every day at the university with great zeal never any longer.’ b. Non seguir6 le lezioni tutti i giorni all’universita mai PIU, diligentemente. c. Non seguiro le lezioni tutti i giorni mai PIU, all’universita, diligentemente. d. Non seguiré le lezioni mai PIU, tutti i giorni, all’ universita, diligentemente.

In other words, when they appear after the complements, “lower” pre-VP AdvPs “close” the VP bearing the nuclear (or focus) stress of the sentence. To summarize, we have arrived at the following overall ordering of adverbs: an ordered sequence of “higher” (sentence) adverbs precedes an ordered sequence of “lower” adverbs that can either appear in front of the VP or at the very end of the VP bearing the nuclear (or focus) stress. In addition to these classes, we observed the existence of various VP-internal postcomplement adverbials that are unordered with respect to one another and precede “lower” AdvPs in the VP-final position (or follow them, if de-accented—yjust like any other type of de-accented material, including “higher” sentence AdvPs): (70) “Higher” (sentence) AdvPs > “Lower” AdvPs > (DP,,,;) (V) complements > Place,time,manner,etc. adverbials > (focused) “Lower” AdvPs > de-accented material

An analogous pattern appears to exist in French. In the next sections, I try to account for the complex distribution of the adverbial classes in postcomplement position, beginning with the second VP-final “space” for “lower” AdvPs.

1.4 Cases of AdvP movement and questions of scope The conclusion reached in the preceding section that there exist two distinct “spaces” for “lower” AdvPs (a pre-VP space and a postcomplement VP-final space), formed by the same adverb classes in the same fixed order, raises the question of how these two “spaces” relate to each other. There are two possibilities. Either the adverbs are independently generated in the two “spaces,” or the two positions are related by movement.” To decide between these two alternatives, let us consider a clear case of AdvP movement and compare it with a clear case of AdvPs independently “base generated” in two (or more) distinct positions. (Certain classes of) AdvPs clearly can move under wh-movement (71) (see, most recently, Kayne 1994, 75; Chomsky 1995, 48 and 390, n. 102), although “higher” AdvPs resist it (72) (see Chomsky 1986, 83; Lonzi 1991, 397):*

ON THE RELATIVE ORDER OF ADVERB PHRASES 17

(71) a. How elegantly do you think he was dressed? b. The harbor was destroyed as completely as they think the town was destroyed. c. How often do you think he will visit us? d. MAT Gianni ti farebbe del male! ‘Never (focus movement) G. would hurt you!’

(72) a. *How luckily has he won? b. *How probably will he arrive late? c. *How courageously has he eaten the mushrooms?

When possible, wh-movement is apparently allowed to subvert the relative order between two AdvPs. So, for example, although gia ‘already’ must precede manner AdvPs (see Tratta gia male il suo assistente “He is already treating his assistant badly’ versus *Tratta male gia il suo assistente “He is badly already treating his assistant’),

it can follow a manner AdvP if this has been wh-moved (Quanto male tratta gia il suo assistente? ‘How badly is he already treating his assistant?’). Gia continues to take scope over the manner AdvP, which suggests that it is the position of the whtrace that matters.*° The wh-moved AdvP also continues to satisfy the subcategorization requirements of the predicate (the preceding trattare and dress in (71a)). The “scope under reconstruction” property and the subcategorization property, in particular, are the hallmarks of the presence of an A-bar chain connecting the AdvP in Spec CP to the position of the trace. But the movement of AdvPs via wh-movement is not really the point. The relevant question is whether an AdvP can move from its “base generation” position to another (non wh-operator) position. Some genuine cases of this sort are reported to exist in the literature. For example, in French, though not in Italian, a number of “lower” AdvPs can apparently move from the “middle field” of (certain) embedded clauses to the “middle field” of the matrix clause (retaining the interpretation and scope of the trace). (See Kayne 1975, 63ff; Pollock 1989, 416; Cinque 1992a.)*’ (73) a. Il (me) faut pas que tu parles. (in the interpretation “JI faut que tu ne parles pas”) ‘It is necessary that you do not speak.’ b. ?Il (ne) faut rien que tu fasses. ‘It is necessary that you do nothing.’ c. Il ne faut plus que tu parles. (in the interpretation “Tl faut que tu ne paries plus”) ‘It is necessary that you don’t speak any longer.’

In this case, too, the AdvPs continue to satisfy the subcategorization requirements of the embedded predicate and retain in their derived position the interpretation associated with their “base” position:** (74) a. ?Vous avez mal dé raccrocher. (Kayne 1975, 27, n. 29) “You must have hung up badly.’ b. If a bien di se comporter. (Cardinaletti and Starke 1994, n. 78) ‘He has well had to behave.’

Again, these properties are indicative of the presence of an A-bar chain connect-

ing the two positions (i.¢e., of a movement relation). Here, however, the relative order of the AdvPs cannot be subverted:

18 ADVERBS AND FUNCTIONAL HEADS

(75) *Ilamal dé toujours raccrocher.” This suggests that order subversion is not an automatic effect of movement, but only of a specific type of movement. A comparison of the preceding wh-movement case and the French case just mentioned suggests that only movement to Spec CP allows for order subversion among adverbs. But the Germanic Verb Second languages provide evidence for a certain qualification of this conclusion. In these languages, movement of an AdvP to Spec CP (to comply with Verb Second requirements) does not automatically allow for subversion of the relative order of two AdvPs. This is explicitly noted in Koster (1978, 205-9), where it is observed that while the rigid relative order of a predicative AP and a sentence AdvP in embedded clauses (see (76)) can be subverted by movement of the AP to first position in matrix clauses (see (77)), the relative order of two AdvPs can never be so subverted (i.e., by moving the lower AdvpP to first position in a matrix clause). Compare (78)and (79): (76) a. Zi zegt dat hij waarschijnlijk erg ziek is. ‘She says that he probably very sick is.’ b. *Zij zegt dat hij erg ziek waarschijnlijk is.

(77) Erg ziek is hij waarschijnlijk niet. ‘Very sick is he probably not.’ (78) a. Het is zo dat hij helaas waarschijnlijk ziek is. ‘It is the case that he unfortunately probably sick is.’ b. *Het is zo dat hij waarschijnlijk helaas ziek 1s.

(79) *Waarschijnlijk is hij helaas ziek.>° (cf. Helaas is hij waarschijnlijk ziek) ‘Probably is he unfortunately sick.’

Koster relates the difference between the predicative AP case (77) and the AdvP case (79) to the fact that movement of the AdvP to first position is apparently not a genuine instance of wh-movement, as witnessed by the fact that no d-proform exists for AdvPs (see (80)) and that no unbounded movement is available to them (see (81)) (unlike the AP case):?!

(80) *Waarschiynlyk dat is hij ziek. ‘Probably that is he sick’ (81) *Waarschijnlijk, zegt Jan dat hij t; ziek is. “Probably says J. that he sick is.’

It is therefore plausible to assimilate the derived position of the AdvP in the COMP “space” of Verb Second languages to a (nonoperator) A-bar position of essentially the same nature as the (nonoperator) A-bar position of “base generation” of AdvPs. This induces a Relativized Minimality (Rizzi 1990) (or “Shortest Movement’; Chomsky 1995) violation whenever another A-bar position of the same type is crossed.°? Similar, though less sharp, judgments are found with the movement of certain “higher” (sentence) AdvPs to the COMP “space” in French. AdvPs that have this possibility are peut-étre, sans doute, and certainement,; “epistemic modal” AdvPs like probablement; and “evaluative” AdvPs like heureusement.>> Sueur (1978, 238) notes contrasts like the following (to be compared with (49a—b), also from Sueur 1978, which showed that in the same “space,” “evaluative” adverbs must precede “epistemic modal” adverbs):

ON THE RELATIVE ORDER OF ADVERB PHRASES 19

(82) a. Heureusement que sans doute, Pierre viendra. ‘Luckily that, undoubtedly, P. will come.’ b. ??Sans doute que heureusement, Pierre viendra. ‘Undoubtedly, luckily, P. will come.’

All this suggests that only movement to an A-bar operator position (wh- or focus) permits subversion of the relative order of two AdvPs. Any other (clear) movement of a nonoperator type does not (plausibly as a consequence of Relativized Minimality). Let us now consider a clear case of independent “base generation” of one AdvP

in distinct positions in the clause. A good example is Jackendoff’s (1972) class I adverbs, which can occupy different positions in a clause with a concomitant change in interpretation: (83) a. John has answered their questions cleverly. b. John cleverly has answered their questions. c. John has cleverly answered their questions. In (83a), cleverly has a pure manner interpretation (the way John answered their questions was clever). In (83b), this interpretation is no longer available, and the adverb

has what he calls a “subject-oriented” interpretation (It was clever of John to have answered their questions). In fact, the two possibilities can be combined, as in (84):°4 (84) John cleverly has answered their questions cleverly/stupidly.

According to Jackendoff, the ambiguity of (83c) is due to the sentence being structurally ambiguous. Because it is adjacent to the main verb, the AdvP can be a VP adverb (hence the availability of the manner interpretation). But the AdvP can also be located outside the VP (possibly in the same position it occupies in (83b), except for the raising of has past it), which gives the equally possible “subject-oriented” interpretation. As expected, if another auxiliary is inserted, the structural ambiguity disappears. See (85a), where only the manner interpretation survives, and (85b), where only the “subject-oriented” interpretation is possible: (85) a. John has been cleverly answering their questions. b. John has cleverly been answering their questions.

It is doubtful, however, that the manner interpretation of the adverb in “auxiliary position” in (85a) is exactly the same manner interpretation of the adverb in (83a). That the two positions of “manner” interpretation should be distinguished is shown by the fact that only one (the postverbal and postobject position), but not the other (the preverbal position) can satisfy subcategorization requirements (see (86)) and can license the middle interpretation (see (87)):> (86) a. John has worded the letter carefully. b. *John has carefully worded the letter.

(87) a. This bag opens up easily. b. *This bag easily opens up. (OK as a plain unaccusative)

Moreover, both can be filled simultaneously: (88) John has been cleverly answering their questions cleverly/stupidly.

20 ADVERBS AND FUNCTIONAL HEADS

These facts clearly indicate that the preverbal “manner” position of the adverb in (85a)

is not transformationally related to the postverbal (and postobject) “manner” position of the adverb in (83a), at least in the same way in which a wh-moved “manner” AdvpP is tranformationally related to the postverbal (and postobject) “manner” position. (Note the systematic contrasts displayed: How carefully has John worded the letter? versus (86b); How easily does this bag open up? versus (87b);* How cleverly has John been answering their questions stupidly? versus (88).) The same conclusion, of course, holds for the “high” position of the adverb in (83b) and (85b) (compare *John carefully has been wording the letter and *John has carefully been wording the letter). This means that there are three positions for the independent “base generation” of the same adverb, which appear to correspond to three different interpretations (even if the difference is obscured in certain contexts, as happens with the preverbal and VP-final positions in (85a) and (83a)). In fact, a restrictive theory should force a one-to-one relation between position and interpretation (1.e., one specific and distinct interpretation for each position of “base generation”). Consequently, whenever one AdvP seems to have exactly the same interpretation in two apparently distinct positions, either (1) 1t occupies the same position and something else has moved around it (as I would claim happens in John

probably has been sick and John has probably been sick),°® (2) it has moved from one position to the other, retaining the interpretation associated with the position of the trace (as in How cleverly has John worded the letter?), or (3) it deceptively has exactly the same interpretation 1n the two positions. Compare (83) and the case of Jackendoff’s class II adverbs (such as guickly and slowly), which seem to have the same interpretation in initial, auxiliary, and VP-final position: (89) a. Slowly(,) John has been dropping his cup of coffee. b. John has been slowly dropping his cup of coffee. c. John has been dropping his cup of coffee slowly.

As Travis (1988, 292ff) and Rochette (1990, 63ff), among others, have observed, the apparently identical interpretation is a misleading impression arising in certain (perhaps, most) contexts. In others, the different interpretations associated with the different positions reemerge. For the pre-VP and VP-final positions, see, for instance, (90a—b), adapted from Thomason and Stalnaker (1973, 200); see also Rochette 1990, 64: (90) a. He has been slowly testing some bulbs. b. He has been testing some bulbs slowly.

In (90a), slowly qualifies (has scope over) the entire event (each test could well have been rapid), whereas in (90b), it qualifies each test individually.*’ Concomitant properties of the second case, where the adverb itself moves (as

opposed to the first and third cases), is that the adverb also retains, in its derived position, the scope and subcategorization properties of the trace position (see Chomsky 1995, 48) and (where the movement is of the unbounded wh-type) the possibility of subverting the rigid relative ordering of the AdvPs. Having compared clear cases of AdvP movement with clear cases of “base generation,” let us go back to the case observed in §1.3 of the “lower” AdvPs appearing

ON THE RELATIVE ORDER OF ADVERB PHRASES 21

in two different “spaces” (one preceding and one following the verb’s complements). Should an AdvP appearing in either of the two “spaces” be “base generated” in one or the other, independently? Or should it be “base generated” in only one of them and “optionally” moved to the other? The properties of the relation seem to suggest a movement derivation (from the lower to the higher “space,” given “proper binding” considerations). First, there is no perceptible change in interpretation between the pre-VP and the VP-final positioning of the AdvPs (if we abstract from focus-presupposition differences). See (91): (91) a. Da allora, non ha mai pia rivisto Maria. ‘Since then, he hasn’t ever any longer seen M.’ b. Da allora, non ha rivisto Maria mai PIU.

Second, there may be a subversion of the rigid relative order of two AdvPs, as already noted. See (92): (92) a. Da allora, non accetta i nostri inviti mica piu sempre. ‘Since then, he doesn’t accept our invitations not any longer always.’ b. Da allora, non accetta sempre i nostri inviti mica PIU. (Cf. *Da allora, non accetta i nostri inviti sempre mica pit.)

Third, the AdvP in the lower “space” takes scope over the AdvP(s) in the higher “space.” Thus, sempre in (92b) 1s understood in the scope of the negative AdvP (mica) PIU (meaning “he no longer always accepts our invitations,” rather than “he always no longer accepts our invitations”). With right adjunction unavailable (Kayne 1994), this can be understood if their relative scope is computed under the reconstruction of sempre, respecting their rigid relative order (mica > pili > sempre). The movement of the AdvP in overt syntax, from the lower to the higher “space,” could be conceived of as motivated by the need to check certain features in the Spec of a higher functional projection (Chomsky 1995) and/or by the morphological “weak” status of the AdvP Gn the spirit of Cardinaletti and Starke 1994). There are, however, certain difficulties with taking an AdvP to move by itself from one “space” to the other. Earlier we noted that subversion of the relative order of two AdvPs appears to be admitted with a certain kind of movement only (wh-, or operator, movement), which 1s also “unbounded.” But the putative movement of AdvPs from the VP-final “space” to the pre-VP “space” does not seem to be of the wh-movement type. The

AdvPs show no wh-modification, nor are they focused, as in Focus movement (a covert case of wh-movement). Moreover, their landing site is not one open to wh-movement, nor is their movement unbounded: the AdvPs cannot be extracted from the clause in which they originate (cf. (93a—b), recalling that di gid precedes completamente in the same “space’’):*8 (93) a. A Natale, credo che avesse completamente perso la testa di GIA. ‘At Christmas, I think he had completely lost his mind already.’ b. *A Natale, credo completamente, che avesse t; perso la testa di GIA. ‘At Christmas, I think completely that he had lost his mind already.’

A possible way to reconcile the observed movement properties of the relation between the two “spaces” with the generalization that subversion of relative order goes

22 ADVERBS AND FUNCTIONAL HEADS

together with unbounded operator movement only is to deny that the AdvP moves by itself and to assume that it moves within a larger constituent.

Consider what the derivation of (93a) would look like under this alternative. From a “base” structure such as (94), in which di gia precedes completamente and the VP, (93a) is derived by raising the constituent including completamente

and whatever follows completamente to the left of di gid, as indicated by the arrow:°?

TH

(94) sempre

> completamente > bene, and if reversals of this order are brought about only by scrambling to the left of a higher AdvP one of the successively larger constituents shown in (104), then only the forms in (105) (with the derivations indicated) are predicted to be possible:

To _ To

(104) Lui non ha mica pit [ xp sempre [ yp completamente | zp rivisto benei suoi

‘He hasn’t any longer always completely corrected well his appunti ]]] notes.’

(105) a. Lui non ha [ zp rivisto bene 1 suoi appunti | mica pit ( xp sempre [ yp

completamente [ zp ] ] ]

Td To

|To

b. Lui non ha mica pit [ zp rivisto bene 1 suoi appunti | [ xp sempre

[ yp completamente [ zp ] ] ]

oo TO

c. Lui non ha mica pit [ xp sempre [ zp rivisto bene i suoi appunti |

_

[ yp completamente [ zp } ] ]

d. Lui non ha [ yp completamente [ zp rivisto bene 1 suoi appunti | ] mica pit

| xp sempre [ yp ] ]

TO

e. Lui non ha mica pit [ yp completamente [ 7p rivisto bene i suoi appunti | ]

_

[ xp Sempre [ yp ] |

f. Lui non ha [ xp sempre [ yp completamente [ zp rivisto bene i suoi appunti | | |

mica piu [ xp]

ON THE RELATIVE ORDER OF ADVERB PHRASES 25

If the “base” structure is (104), scrambling the constituent comprising sempre and everything following it around the higher mica pia must necessarily drag along completamente (which is between sempre and bene) and bene itself. Hence the ungrammaticality of (99a) and (101), where completamente and bene, respectively, have “remained behind.” For the same reason, (99b) has no well-formed derivation. If what moves is the constituent YP of (104), there is no way in which bene could have been left behind. Similar considerations apply to (102). This kind of analysis may shed some light on certain adverbial scope facts noted in the literature but not fully explained.

Recall that in (93a), completamente ‘completely’ is within the scope of di GIA ‘already’ to its right. In the preceding analysis, this was seen as a consequence of di GIA asymmetrically c-commanding the trace (or the copy; Chomsky 1995, chap. 3) of the larger constituent containing completamente, moved across di GIA (scope, here, being computed “under reconstruction’”).© In this light, consider (106a—b), discussed in Andrews (1983). As he notes there, (106a) involves “two instances of intentional knocking” and (106b), “one intentional instance of knocking twice” (p. 695): (106) a. John knocked on the door intentionally twice. b. John knocked on the door twice intentionally.

In each case, the adverb to the right takes scope over the adverb to its left (a “comma intonation” apparently separating the two adverbs; see Fillmore 1994, 170). When

both precede the verb, it is the one to the left that takes scope over the one to the right, as expected. See (107a—b), ignoring for the time being their (different degrees of) marginality noted by Andrews: (107) a. ?John twice intentionally knocked on the door. b. ??John intentionally twice knocked on the door.

To account for the wider scope of twice in (106a) and (107a) and of intentionally in (106b) and (107b), Andrews proposes the following nested V' representations, coupled with the semantic principle “Apply an adverb to what it is sister of” (1983, 695): (108) a. John {y [y [y knocked on the door | intentionally | twice ] b. John [Ly [y [ y: knocked on the door | twice | intentionally | (109) a. ?John [ y: twice [ y: intentionally [ y knocked on the door J | ] b. 7?John [ y: intentionally [ y: twice [ y: knocked on the door | | ]

He also notes that when the two adverbs appear on either side of the most embedded V', the resulting sentences (110a—b) are ambiguous (as predicted by the nested V' theory, for “an Adv—V'—Adv sequence will be able to be structured either as [ Adv [y V' Adv ]]] or as [y[Ly Adv V’ ] Adv ]” (p. 696): (110) a. John intentionally knocked on the door twice. b. John twice knocked on the door intentionaily.

Attractive as it may seem, this analysis raises some questions. First, as Andrews himself admits, 1t provides “no explanation for the apparent predominance of the [106a] reading for [1 10b]’” (1983, 696) (namely, for the predominantly wide scope reading of twice in (110b)).° Second, if Kayne’s (1994) antisymmetry theory is correct, the rep-

26 ADVERBS AND FUNCTIONAL HEADS

resentations (108a—b) are problematic as “base generated” representations of the relative scope of twice and intentionally, as they involve right adjunctions. From that point of view, the scope facts of (106a—b) suggest, if anything, that the first of the two adverbs has raised, within a larger phrase, across the second, which takes wider scope as a consequence of being generated higher (just like the case of di GIA in (93)). It is thus tempting to derive (106a) from the (virtually) unambiguous (110b) via movement of the constituent [knocked on the door intentionally] around twice, as shown in (111) (whence the characteristic “comma intonation” before twice, noted by Fillmore 1994—-corresponding to that before di GIA in (93a)—and the wider scope of twice):

TO

fT

(114) John twice [ knocked on the door intentionally |

By the same token, (106b) must be derived from (110a) by moving the constituent [knocked on the door twice] around intentionally: (112) John intentionally [ knocked on the door twice |

ee

We should be more precise, though, since (110a) is, in fact, ambiguous. If we take the representations in (113a—b) to correspond to the two readings of (110a), then (106b) is more accurately derived from (113a), as was shown in (112): (113) a. John [ intentionally [ knocked on the door twice ] ] b. John [ intentionally [ knocked on the door | ] [twice t ]

This seems to have led us to a paradox, since we have postulated both the structure (111) with twice higher than intentionally and the structure (112) with twice lower than intentionally. Example (111) underlies the (virtually) unambiguous (110b); (112) undelies (106b) with the movement indicated and one of the two interpretations of (110a) if the movement indicated does not take place. The paradox, however, is not real, as there 1s evidence that twice belongs to a class of adverbs (many/few/etc. times, often, rarely, frequently, etc.) that are systematically ambiguous between two interpretations, each associated with a different position. The higher position quantifies over the entire event (saying how frequently it takes place). In (111), for example, it says that there were two events of knocking on the door (intentionally). The lower position, instead, just indicates the repetition of the act denoted by the verb. So, (1 13a) says that there was a single event of (intentional) repetition of the act of knocking on the door.® The different semantics associated with the two positions also shows up in a case like the following (with often replacing twice, to make the judgment sharper): (114) a. Texans often drink beer. b. Texans drink beer often.

The higher often of (114a), though not the lower one of (114b), may act as an “adverb of quantification” (in the sense of Lewis 1975). It is able to unselectively bind the bare DP subject inducing the interpretation “Most Texans drink beer” (without saying how frequently they drink it).”° Often in (114b), instead, cannot bind the bare DP subject (which is, rather, bound by a generic operator—Chierchia forthcoming—

ON THE RELATIVE ORDER OF ADVERB PHRASES 27

roughly meaning “typically all’’). Rather, it says that beer drinking takes place more times than is usually the norm.” A clear indication that there are indeed two distinct positions is the fact that they can be simultaneously filled. In this case, the higher one unambiguously quantifies over the event; the lower over the act:”2 (115) a. John twice (often/rarely/ . . .) knocked on the door twice (three times/often/ . . .). b. John twice (often/rarely/ .. .) knocked twice (three times/often/ . . .) on the door.

We may now ask whether having recognized the existence of two different positions for twice provides a full account of Andrews’s observations. The derivations (111)

and (112) just sketched suggest that not only twice but also intentionally must be able to occur in two distinct positions. The latter conjecture, however, does not seem empirically well motivated. First, the sentence John intentionally knocked on the door intentionally makes little sense in contrast to the perfectly sensible John twice knocked on the door twice.

Second, the right core orders and scopes appear to be derivable from the single underlying structure (116), with one position for intentionally and the observed two positions (and interpretations) for twice: (116) John ( twice, ) [ xp intentionally [ yp knocked ( twice, ) on the door | |

If twice, is absent and nothing moves, (107a) is obtained, with (frequentative) twice taking scope over intentionally. As for the marginality of (107b), I take it to be due to the low acceptability of a sentence initial (“adverb of setting’) position of intentionally, to the left of twice,: ??Intentionally John twice knocked on the door, from which (107b) is derived by moving the subject across intentionally (If 7 John has intentionally twice knocked on the door twice is approximately of the same acceptability level (27), then the initial adverb must also be crossed over by the auxiliary). The interpretation of (110a) involving “one intentional instance of knocking twice” is derived from (116) if twice,, but not twice), is present and nothing moves except on the door across twice, (this should correlate with the absence of a “comma intonation” before twice). The other interpretation of (110a)—that involving “two instances of intentional knocking”—1s instead derived from (116) if twice,, but not twice>, is present and XP moves around twice, (this reading should correlate with a “comma intonation” before twice).”3 Example (106a) is instead derived if twice,, but not twice, is present, and, first, YP moves around intentionally and then the resulting XP (knocked on the door intentionally) moves around twice. Here, the expectation 1s that intentionally continues to take scope over knocked on the door and that twice takes scope over knocked on the door intentionally and is separated from it by a “comma intonation.” Iftwice, were also present, it would then appear either after or before the PP on the door. See (117a-b): (117) a. John knocked twice on the door intentionally twice. b. John knocked on the door twice intentionally twice.

Example (106b) is derived if twice,, but not twice,, is present, on the door moves around it, and YP moves around intentionally, which should then be preceded by a

28 ADVERBS AND FUNCTIONAL HEADS

“comma intonation.” If twice, were also present, we would have either (118a) or (118b), according to where YP (of (116)) moves: (118) a. John twice [{ knocked on the door twice ] intentionally. b. John [ knocked on the door twice } twice intentionally.

Also note that the virtual nonambiguity of (110b), with the predominance of the wide scope reading of twice (mysterious in Andrews’s analysis), becomes understandable under the present analysis. The predominant reading 1s derived from (116), with twice, but not twice,, present and with movement of YP around intentionally.“ The other reading, with intentionally taking scope over twice is not derivable from (116). It can be derived only from a structure like (119), in which intentionally is higher than twice,;, by moving XP around intentionally. But, as Andrews already observed, the order intentionally > twice, is very marginal.” (119) John intentionally [ yp ( twice, ) [ yp knocked ( twice, ) on the door ] |]

1.5 Circumstantial adverbials of place, time, manner, and the like In this section I briefly consider the class of adverbials, sometimes called circumstantial (see Ruwet 1968, 353; Lonz1 1991, 381; Fillmore 1994, among others), which follow the verb’s complements within the VP. They comprise a varied selection of elements: place, time, manner, means, company, reason, purpose, and so forth. As noted in §1.3, they appear to differ from the adverb classes just considered (the AdvPs proper) in not being rigidly ordered with respect to one another. See Chomsky (1995, 333) and p. 15f. In contrast to AdvPs proper, they can also be interchangeably in one another’s scope, depending on their mutual structural relation. In (120a), the place adverbial (containing an indefinite DP) is in the scope of the time adverbial containing a universal quantifier. In (120b), it 1s the other way around: (120) a. He attended classes every day of the week in a different university. b. He attended classes in each university on a different day of the week.

A similar case is in (121), with place and company PP adverbials: (121) a. He attended classes in each university with a different friend. b. He attended classes with each friend in a different university.

Circumstantial adverbials also differ from AdvPs proper in that they are typically realized (with the partial exception of manner adverbials) in prepositional form (for three hours, in the kitchen, with great zeal, for your love, in a rude manner, with a bicycle, etc.) or in bare NP form (the day after, tomorrow, this way, here, etc.; see Larson 1985, Stroik 1992). Furthermore, possibly as a consequence of this, they cannot appear in any of the pre-VP positions open to AdvPs proper (except for the absolute initial position of “adverbs of setting,” a topic-like position).”® Finally, as pointed out to me by @ystein Nilsen, they also appear to differ semantically from AdvPs proper. Whereas the latter are characteristically operators (functions mapping propositions to propositions, or predicates to predicates), circumstantial adverbials can be seen (after Davidson 1967) as modifiers predicated of an underlying event variable. For semantic arguments to this effect, see Parsons (1990,

ON THE RELATIVE ORDER OF ADVERB PHRASES 29

chap. 4), who also treats the preverbal class of manner adverbs in the same way. These five properties, distinguishing circumstantial adverbials from the AdvPs proper examined here, suggest an entirely separate treatment for them. If AdvPs proper occupy the specifier position of distinct functional projections above the VP, as I argue in the following chapters, it seems natural not to assume the same for the circumstantial phrases. This is particularly natural if the rigid ordering

of AdvPs is a consequence (via Spec/head agreement) of the rigid ordering of the respective functional heads (see chapter 4). In this view, the “free” order of circumstantial phrases would correlate with the fact that they are not generated in specifier positions of functional projections: something rendered independently plausible by the standard view that functional projections are “shells” external to the lexical projection, here VP, not its most deeply embedded ones. If so, what is the structural position of circumstantial phrases? One possibility would be to follow Chomsky’s (1995, 333) tentative conclusion that “if a shell structure is relevant at all, the additional phrases might be supported by empty heads below the main verb.’”””” Another very interesting possibility is suggested by Mystein Nilsen (1998) based on a parallelism between the structure of clauses and that of DPs. Just as DPs have, in addition to attributive adjectives (in pre-NP position), predicative adjectives (small, or reduced relative, clauses) as their most deeply embedded constituents in the NP (Cinque 1994, Kayne 1994, §8.4), so clauses should have “attributive” adverbials (in pre-VP position) and “predicative” adverbials as their most deeply embedded constituents in the VP. Taking the parailelism strictly implies, according to Nilsen, that the postcomplement circumstantial adverbials are actually “predicates” predicated of the VP (in a way reminiscent of the semantic treatment of circumstantial adverbials in the Davidsonian tradition mentioned earlier).’? This would

amount to assigning to a sentence such as (122) a structure like (123), where at the university is predicated of the VP John attended classes, and every day is predicated of the larger VP John attended classes at the university. (122) John attended classes at the university every day. (123) F;

F, F; VP

a VP every day

VP 4S»

30 ADVERBS AND FUNCTIONAL HEADS

A variant of this idea is to take a configuration such as (123) not as “base generated” but as derived from an underlying structure like (124), with the adverbial PPs in the spec of distinct VP “shells,” followed by obligatory successive leftward movements of the lower VPs to higher specifiers (perhaps to establish the required predication): (124) F,

F, F, VP,

every day =v IN at the univ vy sSOV*P,

John attended classes

The latter analysis would, in fact, seem appropriate to account for the scope facts of (120) and (121) and for the stranding of PPs and relative clauses in final or intermediate positions in the (layered) VP. I refer to Nilsen (1998) for further discussion.

1.6 “Focusing” and “parenthetical” uses of AdvPs The aim of this section is not to discuss “focusing” and “parenthetical” uses of adverbs in general but, rather, to expose certain potential sources of counterevidence to the ordering of AdvPs examined earlier, arguing for their irrelevance. In note 12, I suggested that certain unexpected orders of gia ‘already’ in Italian, and its Norwegian equivalent allerede, could be understood as instances of a “focusing” usage of the adverb. By that I meant—perhaps slightly extending the term— that the adverb might modify directly different types of constituents (to account for its apparent multiple positioning). This is a prototypical property of such (exclusively) “focusing” adverbs as only and even, which are found preceding DPs, APs, AdvPs, PPs, and VPs, as well as various clausal functional projections (Rooth 1985; Bayer

1996, 11ff):” (125) a. He likes only himself. b. The success was only partial. c. He solved the problem only partially. d. He will have been beaten only by John. e. He will have been only beaten by John.

ON THE RELATIVE ORDER OF ADVERB PHRASES 31

f. He will have only been beaten by John. g. He will only have been beaten by John. h. ?He only will have been beaten by John. i. He said only that he doesn’t like it.

Just as with gid, we find that many more adverbs have such a use (hence certain unexpected orders). So, for example, there is evidence that most classes of “higher” AdvPs also have this use. “Higher” AdvPs in Italian, as a rule, cannot occur after a finite verb (in a sentence-final position, unless they are “de-accented’’):®° (126) a. *Gianni lo merita francamente/fortunatamente/evidentemente/probabilmente/

forse/... *G. deserves it frankly/luckily/evidently/probably/perhaps/ .. .’ b. Gianni lo merita, francamente/fortunatamente/evidentemente/probabilmente/

forse/...

The same position, however, becomes immediately available if another constituent follows the adverb, in which case, as noted in Belletti (1990, 130, n. 29), the adverb takes scope not over the sentence but just over that constituent:*! (127) Gianni lo merita francamente/fortunatamente/evidentemente/probabilmente/forse/ ... per pil di una ragione. ‘G. deserves it frankly/luckily/evidently/probably/perhaps/ . . . for more than one reason.’

Additional evidence for the conclusion that they indeed can form a constituent with the phrase following them 1s the fact that they can be fronted together under Focus Movement, or Clefting (see (128)) and the fact that they can intervene between a verb and a (light) object (see (129)), a possibility not open to “nonfocusing” adverbs (T. B. Ernst 1984, 111ff; 1991, 752ff; Rooth 1985) (see (130)):*2 (128) a. Probabilmente per questa RAGIONE, lo hanno licenziato. ‘Probably for this reason (focus), they have fired him.’ b. E’ probabilmente per questa ragione che lo hanno licenziato. ‘It is probably for this reason that they fired him.’ c. C’est heureusement Paul qui a vendu sa voiture.*° “It’s luckily P. who sold his car.’

(129) a. He hates probably everybody. b. He has drunk already seven beers. (130) *He forgot often his name.%*

Assuming a simple adjunction to a maximal projection to be excluded (Kayne 1994), itis plausible to treat “focusing” adverbs as heads taking their modifees as comple-

ments (see Bayer 1996, chap. 1). This appears natural in those cases where the “focusing” adverb immediately precedes the phrase in focus. Such alternations as (131a—b) and a comparison between (132) and (133) seem to suggest that the complement of solo ‘only’ and too raises to the Spec, across the head:®°

(131) a. Solo MARIA hanno visto. ‘Only M. (focus) they saw.’ b. Maria SOLO hanno visto. ‘Mary only (focus) they saw.’

32 ADVERBS AND FUNCTIONAL HEADS

(132) Even Mary was available. (133) Mary too was available. What is relevant here is that the unexpected ordering of “higher” adverbs after “lower” ones in such contexts as (134a—c) can plausibly be seen as a consequence of their “focusing” usage: (134) a. Lo avra gia detto probabilmente a tutti. ‘He will have already said that probably to everybody.’ b. Ha sempre presentato purtroppo la stessa versione. ‘He always presented unfortunately the same version.’ c. Non legge piu romanzi forse proprio per questo. ‘He doesn’t any longer read novels perhaps just for this reason.’

Besides gid, other “lower” adverbs also have “focusing” uses: one example ts ancora

= ‘again’ and, more marginally, the ancora of non... ancora ‘not yet’ (though not the ancora meaning ‘still’) (see (135a—c)). Analogously, spesso ‘often’ can be used as a “focusing” adverb (see (136)): (135) a. Visitera (ancora) Maria a Roma (ancora) due volte. ‘He will visit (again) M. in Rome (again) twice.’ b. Gianni non ha (ancora) ricevuto biglietti di auguri (?ancora) da nessuno. ‘G. hasn’t (yet) received cards (yet) from anybody.’ c. Gianni manda (ancora) biglietti di auguri (*ancora) agli amici. ‘G. sends (still) cards (still) to his friends.’ (136) Gianni ha mandato i biglietti di auguri agli amici spesso in ritardo. ‘G. sent his cards to his friends often late.’

A more systematic survey is beyond the scope (and the aims) of this work.* Finally, I should mention that another potential source of unexpected orders is the “parenthetical” usage of (typically “higher’’) adverbs, something for which I have

no interesting account to propose. As shown in (137), such “higher” adverbs as purtroppo ‘unfortunately’, forse ‘perhaps’, and francamente ‘frankly’ are possible even after “lower” adverbs if set off by “comma intonation” from the rest:®’ (137) a. Da allora, non é piu, purtroppo, venuto a trovarcli. ‘Since then, he has no longer, unfortunately, come to visit us.’ b. Riusciro a leggere tutto, forse, per la prossima settimana. ‘I will manage to read everything, perhaps, for the next week.’ c. Lui mi é sempre parso, francamente, una persona difficile. ‘He always seemed to me, frankly, a difficult person.’

1.7 Toward a universal hierarchy of AdvPs: some cross-linguistic evidence In this section, I consider some other languages (and language families) to see if the conclusions on the order of AdvPs reached on the basis of Romance have more general validity. Although by the logic of the inquiry, we cannot be certain, the

ON THE RELATIVE ORDER OF ADVERB PHRASES 33

striking consistency found in the languages reviewed here makes me claim that a universal order, or hierarchy, of AdvPs does exist, in spite of the questions that remain to be clarified (and of the problems that a closer look will almost certainly reveal), I consider two Germanic languages (English and Norwegian), a Slavic language (Bosnian/Serbo-Croatian), a Semitic language (Hebrew), a Sino-Tibetan language (Chinese), Albanian, and, more briefly, an Austronesian language (Malagasy). I have kept my commentary to a minimum, in most cases just listing the crucial pairs of sentences that point to a certain relative order between two (or more) adverbs.

1.7.1 English The literature cited in §1.2 shows that English “higher” (sentence) adverbs are ordered as in Romance. Speech act adverbs precede evaluative adverbs (Honestly I am unfortunately unable to help you versus * Unfortunately I am honestly unable to help you). Evaluative adverbs precede evidential adverbs (Fortunately, he had evidently had his own opinion of the matter versus *Evidently he had fortunately had his own opinion of the matter; Siewierska 1992, 418). Evidential adverbs precede epistemic adverbs (Clearly John probably will quickly learn French perfectly versus * Probably John clearly will quickly learn French perfectly; Bowers 1993, n. 13 and p. 607). Epistemic adverbs precede (past) tense adverbs (Probably he once had a better opinion of us), although they can also follow them when they are generated in a higher “scene setting” position (see §1.2: Once he probably had a better opinion of us). Tense adverbs precede perhaps/(almost) certainly (He was then almost certainly/ perhaps at home versus “He was almost certainly/perhaps then at home).®8 Perhaps precedes subject-oriented adverbs (John will perhaps wisely withdraw versus * ?John will wisely perhaps withdraw). Concerning the order of “lower” adverbs, some words are required, as English seemingly differs from Romance on certain points. For example, in Italian (Romance)

(non...) pia “(not ...) any longer’ was seen to precede sempre ‘always’, whereas English often shows the opposite order: (138) John doesn’t always win any longer.

The appearance is deceiving, however. When it follows always, (not...) any longer necessarily follows the verb and its objects, if any, just as is possible in Italian when pia is emphasized: (139) a. John doesn’t always win his games any longer. b. J. non vince sempre le sue partite, PIU.

[ assume that (139a) is derived in the same way as the Italian cases discussed in §1.4, through movement of some projection containing always, the verb, and its complements, to the left of any longer. In fact, when both always and (not .. .) any longer appear before the verb, their order is just like that in Italian (Romance):®° (140) a. John doesn’t any longer always win his games. b. *John doesn’t always any longer win his games.

34 ADVERBS AND FUNCTIONAL HEADS

(141) a. Gianni non vince pin sempre le sue partite. b. *Gianni non vince sempre pitt le sue partite.

A similar situation is found with other lower adverbs (see, e.g., He hasn’t completely ruined it yet, He hasn’t yet completely ruined it versus *He hasn’t completely yet ruined it).

Once these and similar cases are clarified, the order of “lower” adverbs in English does appear comparable to that found in Romance, with habitual adverbs preceding already, which precedes no longer/not . . .) any longer, which precedes always, which precedes completely, which precedes well (and other manner adverbs). In sum, the order of the various classes of AdvPs in English corresponds to that found in Romance and includes the following: (142) frankly > fortunately > allegedly > probably > once/then > perhaps > wisely > usually > already > no longer > always > completely > well

1.7.2 Norwegian erlig talt ‘honestly’ > heldigvis ‘fortunately’ (143) a. Per forlater erlig talt heldigvis na selskapet.”” ‘Peter leaves honestly spoken fortunately now the party.’ b. *Per forlater heldigvis erlig talt na selskapet. ‘Peter leaves fortunately honestly spoken now the party.’

heldigvis ‘fortunately’ > tydeligvis ‘evidently’ (144) a. Per har heldigvis tydeligvis gatt. ‘Peter has fortunately evidently left.’ b. ??Per har tydeligvis heldigvis gatt. ‘Peter has evidently fortunately left.’

tydeligvis ‘evidently’ > sannsynligvis ‘probably’ (145) a. Per skjgnner tydeligvis sannsynligvis problemet godt. ‘Peter understands evidently probably the problem well.’ b. *Per skj@nner sannsynligvis tydeligvis problemet godt. ‘Peter understands probably evidently the problem well.’

sannsynligvis ‘probably’ > na ‘now’ (146) a. Per forlater heldigvis sannsynligvis nd selskapet.”! ‘Peter leaves fortunately probably now the party.’ b. Per forlater heldigvis nad sannsynligvis selskapet. ‘Peter leaves fortunately now probably the party.’

na ‘now’ > kanskje ‘perhaps.’ (147) a. Haner n@ kanskje hjemme. ‘He is now perhaps at home.’ b. *Han er kanskje nd hjemme. ‘He is perhaps now at home.’

ON THE RELATIVE ORDER OF ADVERB PHRASES 35

kanskje ‘perhaps’ > klokelig ‘wisely’ (148) a. Per ble kanskje klokelig allerede sent hjem klokken ti. ‘P. was maybe wisely already sent home at ten o’clock.’ b. *Per ble klokelig kanskje allerede sent hjem klokken ti. ‘P. was wisely maybe already sent home at ten o’clock.’

klokelig ‘wisely’ > vanligvis ‘usually’ (149) a. Klokken to har Per klokelig vanligvis allerede spist. ‘At two o’clock has P. wisely usually already eaten.’ b. *Klokken to har Per vanligvis klokelig allerede spist. ‘At two o’clock has P. usually wisely already eaten.’

vanligvis ‘usually’ > allerede ‘already’ (150) a. Klokken to har Per vanligvis allerede spist. ‘At two o’clock has P. usually already eaten.’ b. *Klokken to har Per allerede vanligvis spist. ‘At two o’clock has P. already usually eaten.’

allerede ‘already’ > ikke lenger ‘no longer’ (151) a. Per elsker allerede ikke lenger sin kone.” ‘P. loves already no longer his wife.’ b. *Per elsker ikke lenger allerede sin kone.” “P. loves no longer already his wife.’

ikke lenger ‘no longer’ > alltid ‘always’ (152) a. Jon aksepterer ikke lenger alitid var invitasjon. ‘J. accepts no longer always our invitation.’ b. *Jon aksepterer alltid ikke lenger var invitasjon. ‘J. accepts always no longer our invitation.’ c. *Jon aksepterer ikke alltid lenger var invitasjon. ‘J. accepts no always longer our invitation.’

alltid ‘always’ > helt ‘completely’ (153) a. De forstar enda ikke alltid helt hva jeg snakker om. ‘They understand still not always completely what I talk about.’ “They do not yet always completely understand what I talk about.’ b. *De forstar enda ikke helt alltid hva jeg snakker om. “They understand still not completely always what I talk about.’ ‘They do not yet completely always understand what I talk about.’

helt ‘completely’ > godt ‘well’ (154) a. ...at han ikke lenger alltid helt sov (sa) godt. ‘that he not any longer always completely slept (so) well.’ b. *... at han ikke lenger alltid sov (sa) godt helt. ‘that he not any longer always slept (so) well completely.’

36 ADVERBS AND FUNCTIONAL HEADS

The following is the relative order of the main adverb classes of Norwegian (see Nilsen 1997, 1998 for a discussion of other classes): (155) erlig talt ‘honestly’ > heldigvis ‘fortunately’ > tydeligvis ‘evidently’ > sannsynligvis ‘probably’ > nad ‘now’ > kanskje ‘perhaps’ > klokelig ‘wisely’ > vanligvis ‘usually’ > allerede ‘already’ > ikke lenger ‘no longer’ > alltid ‘always’ > helt ‘completely’ > godt ‘well’

1.7.3. Bosnian/Serbo-Croatian iskreno ‘frankly’ > naZalost ‘unfortunately’ (156) a. Iskreno, ja nazalost imam jako loSe mi8ljenje o vama.” ‘Frankly, I unfortunately have a very bad opinion of you.’ b. *NaZalost, ja iskreno imam jako loSe miSljenje o vama.

nazalost ‘unfortunately’ > ocigledno ‘evidently’ (157) a. Ja naZzalost ocigledno imam potpuno pogreSnu predstavu o vama. ‘I unfortunately evidently have a completely wrong perception about you.’ b. *Ja ocigledno naZalost imam potpuno pogresnu predstavu o vama.

ocigledno ‘evidently’ > mozda ‘perhaps’ (158) a. On je ocigledno mo2da pretjerao u svojim izrazima zahvalnosti.”° ‘He is evidently perhaps gone too far in his expressions (of) gratitude.’ b. *On je mozda ocigledno pretjerao u svojim izrazima zahvalnosti.

neizostavno ‘necessarily’ > inteligentno ‘intelligently’ (159) a. On ée neizostavno inteligentno odgovoriti na svako neprijatno pitanje. ‘He will necessarily intelligently answer every unpleasant question.’ b. *On ¢e inteligentno neizostavno odgovoriti na svako neprijatno pitanje.

inteligentno ‘intelligently’ > obicno ‘usually’ (160) a. Inteligentno, on obicno nikad ne odgovori na neprijatna pitanja. ‘Intelligently, he usually never answers unpleasant questions.’ b. *Obicno, on inteligentno nikad ne odgovori na neprijatna pitanja.”®

obiéno ‘usually’ > éesto ‘often’ (161) a. On obicno esto navraéa vu moju Kancelariju. ‘He usually often comes in my office.’ b. *On éesto obicno navraéa u moju Kancelariju.

Cesto ‘often’ > veé ‘already’ (162) a. On cesto veé rano ujutro navrati u moju kancelaryu. ‘He often already early in morning comes in my office.’ b. *On veé esto rano ujutro navrati u moyu Kancelariju.

ON THE RELATIVE ORDER OF ADVERB PHRASES 37

veé ‘already’ > vise ‘no longer’ (163) a. U to vrijeme nije ved vise niSta imala. in that time negAUX already no longer nothing have-FEM ‘At that time, she already did not have anything any longer.’ b. *U to vrijeme nije vite veé niSta imala.

vi§e ‘no longer’ > uvijek ‘always’ (164) a. On vise uvijek ne pobjedjuje.”’ he no longer always not win-3SG. b. *On uvijek vise ne pobjedjuje.

uvijek ‘always’ > upravo ‘just’ (165) a. Kad god ga sretnem, on se uvijek upravo vraéa iz grada. “Whenever him I-meet, he REFL always just returns from town.’ b. *Kad god ga sretnem, on se upravo uvijek vraéa iz grada. “Whenever him I-meet, he REFL just always returns from town.’

upravo ‘just’ > gotovo ‘almost’ (166) a. Jasam upravo gotovo pao. ‘T have (Jit. ‘am’) just almost fallen.’

b. *Ja sam gotovo upravo pao. ‘T have (lit. “‘am’) almost just fallen.’

gotovo ‘almost’ > potpuno ‘completely’ (167) a. Jasam ga gotovo potpuno zaboravio. ‘[ have him almost completely forgot.’ b. *Ja sam ga potpuno gotovo zaboravio. ‘I have him completely almost forgot.’

potpuno ‘completely’ > dobro ‘well’ (168) a. Potpuno dobro shvatam tvoje razloge za odbijanje te ponude.”® ‘I completely well understand your reasons for rejecting that offer.’ b. *Dobro potpuno shvatam tvoje razloge za odbijanje te ponude.

brzo ‘quickly’ > opet ‘again’ (169) a. Brzo opet ponovi Sta si rekao. ‘Quickly again repeat what (you) said.’ b. *Opet brzo ponovi Sta si rekao.

The overall order thus seems to be: (170) iskreno ‘frankly’ > naZalost ‘unfortunately’ > ocigledno ‘evidently’ > mozda ‘perhaps’ > neizostavno ‘necessarily’ > inteligentno ‘intelligently’ > obicno ‘usually’ > Cesto ‘often’ > veé ‘already’ > vi¥e ‘no longer’ > uvijek ‘always’ > upravo ‘just’ > gotovo ‘almost’ > potpuno ‘completely’ > dobro ‘well’ > brzo ‘quickly’ > opet ‘again’

38 ADVERBS AND FUNCTIONAL HEADS

1.7.4 Hebrew speech act adverb > evaluative adverb (171) a. Dani be’emet le-mazal-o ha-ra hifsid ‘et ha-rakevet.?° D. in-truth in-luck-his the-bad missed ACC the-train ‘D. in earnest unluckily missed the train.’ b. *Dani le-mazal-o ha-ra be’emet hifsid ‘et ha-rakevet.!°° D. in-luck-his the-bad in-truth missed ACC the-train ‘D. unluckily in earnest missed the train.’

evaluative adverb > epistemic adverb (172) a. Le mazal-a kanir’e Rina tenaceax. luckily apparently R. win-fut ‘Luckily probably R. will win.’ b. *Kanir’e le mazal-a Rina tenaceax.'°! apparently luckily R. win-fut ‘Probably luckily R. will win.’

epistemic adverb > ‘perhaps’ (173) a. Hu kanir’e ‘ulay yaskim. he apparently perhaps agree-fut ‘Probably he will perhaps agree.’ b. *Hu ‘ulay kanir’e yaskim.'” he perhaps apparently agree-fut ‘Perhaps he will probably agree.’

‘perhaps’ > subject-oriented adverb (174) a. Dani ‘ulay be-xoxma yitpater. D. perhaps with-intelligence will resign ‘D. perhaps will intelligently resign.’ b. *Dani be-xoxma ‘ulay yitpater. D. with-intelligence perhaps will resign

epistemic adverb > habitual adverb (175) a. Hu kanir’e be-derex klal me’ axer. he apparently usually late ‘Probably he is usually late.’ b. *Hu be-derex klal kanir’e me’ axer. he usually apparently late ‘Usually he is probably late.’

kvar ‘already’ > tamid ‘always’ (176) a. Hu kvar tamid yodea ‘et ha-tSuva. ‘He already always knows the answer,’ b. “Hu tamid kvar yodea ‘et ha-tSuva.

ON THE RELATIVE ORDER OF ADVERB PHRASES 39

kvar lo ‘no longer’ > tamid ‘always’ '®° (177) a. Hu kvar lo tamid yodea ‘et ha-tsuva. he already not always knows the answer ‘He no longer always knows the answer.’ b. *Hu tamid kvar lo yodea ‘et ha-tSuva. ‘He always no longer knows the answer.’

tamid ‘always’ > bidiuk ‘just’ (178) a. (kSe-‘ani poges ‘oto) John tamid bidiuk xozer me-xul. ‘(When I meet him) J. always just return from abroad.’ b. *(kSe-‘ani poges ‘oto) J. bidiuk tamid xozer me-xul.

tamid ‘always’ > legamrey ‘completely’ (179) a. Hu tamid hores legamrey ‘et ma $e-hu ose. ‘He always destroys completely what he does.’ b. *Hu hores legamrey tamid ‘et ma Se-hu ose. ‘He completely always destroys what he does.’

kim’at ‘almost’ > legamrey ‘completely’ (180) a. John kim’at, legamrey haras ’et ha-misxak. ‘J. almost, completely destroyed the game.’ b. *John legamrey kim’at haras ’et ha-misxak. ‘J. completely almost destroyed the game.’

legamrey ‘completely’ > heitev ‘well’ (181) a. Hu tamid hevin legamrey heitev ‘et dvarexa.'™ ‘He always understands completely well your words.’ b. *Hu tamid hevin heitev legamrey ‘et dvarexa.

The hierarchy thus seems to be: (182) be’emet ‘truly’ > le-mazal-o ha-ra “unluckily’ > kanir’e ‘probably’ > ’ulay ‘perhaps’ > be-xoxma ‘intelligently’ > be-derex klal ‘usually’ > kvar ‘already’ / kvar lo ‘no longer’ > tamid ‘always’ > bidiuk ‘just’ / kim’at ‘almost’ > legamrey ‘completely > heitev ‘well’

1.7.5 Chinese laoshi-shuo ‘honestly’ > buxing ‘unfortunately’ (183) a. laoshi-shuo wo buxing dui tamen you pian-jian. (H)!%° honestly I unfortunately to them have prejudice ‘Honestly I unfortunately have prejudice against them.’ b. *buxing wo laoshi-shuo dui tamen you pian-jian.

40 ADVERBS AND FUNCTIONAL HEADS

buxing ‘unfortunately’ > xianran ‘evidently’ (184) a. buxing tamen xianran dui ni wuhui hen shen. (H)! unfortunately they evidently to you misunderstanding very deep ‘Unfortunately they evidently have a deep misunderstanding of you.’ b. ?*xianran tamen buxing dui ni wuhui hen shen.

xtanran ‘evidently’ > xianzai ‘now’ (185) a. xianran xianzai ni yinggai qu le. (C) evidently now you must go PERF b. ?xianzai xianran ni yinggai qu le.

xianzai ‘now’ > yexu ‘perhaps’ (186) a. Ta xianzai yexu qicheng le. (C) he now perhaps left PERF ‘He has now perhaps left.’ b. Ta yexu xianzai qicheng le.

yexu ‘perhaps’ > mingzhide ‘wisely’ (187) a. Ta yexu mingzhide likai le. (X)!° he perhaps wisely left PERF b. *Ta mingzhide yexu likai le.

mingzhide ‘wisely’ > yiban ‘usually’ (188) a. Ta mingzhide yiban bu biaotai. (X)!°8 ‘He wisely usually does not commit himself.’ b. Ta yiban mingzhide bu biaotai.

yiban ‘usually’ > changchang ‘often’ (189) a. xiatian wo yiban changchang qu haibian. (C) ‘In the summer I usually often go to the sea.’ b. *xiatian wo changchang yiban qu haibian.

changchang ‘often’ > yijing ‘already’ (190) a. Mei tian ba dian, ta changchang yijing chi guo fan le. (C)'”’ every day 8 hours, he often already eat ASP food ‘Every day at 8, he often has already eaten.’ b. *Mei tian ba dian, ta yijing changchang chi guo fan le.

yijing ‘already’ > bu-zai ‘no longer’ (191) a. Ta yijing bu-zai xiangnian ta. (H) he already not-again misses him/her ‘He already no longer misses her.’ b. *Ta bu-zai yijing xiangnian ta.

ON THE RELATIVE ORDER OF ADVERB PHRASES 41

bu-zai ‘no longer’ > zongshi ‘always’ (192) a. ta bu-zai zongshi gen da-ge zhengcao. (H) he not-again always with big-brother quarrel ‘He no longer always quarrels with Big Brother.’ b. *ta zongshi bu~-zai gen da-ge zhengcao.

zongshi ‘always’ > yizhi ‘continuously’ (193) a. Zheyang de wenti, women zongshi yizhi yao taolun liang ge xiaoshi. (C)'!° such DE question, we always continuously must discuss two GE hours b. ??Zheyang de wenti, women yizhi zongshi yao taolun liang ge xiaoshi.

zongshi ‘always’ > ganggang ‘just’ (194) a. mei ci wo pengyjian ta, ta zongshi ganggang cong guowai huilai. (H) every time I meet him, he always just from abroad return ‘Every time I meet him, he has always just returned from abroad.’ b. *mei ci wo pengjian ta, ta ganggang zongshi cong guowai huilai.

ganggang ‘just’? > wanquan ‘completely’ (195) a. wo ganggang wanquan wang-le ta-de dizhi. (H) [ just completely forgot his address ‘I have just completely forgotten his address.’ b. *wo wanquan ganggang wang-le ta-de dizhi.

wanquan ‘completely’ > hao ‘well’ (196) a. ta wanquan gaixie de hen hao. (H) he completely revise till very well ‘He completely revised it very well.’ b. *ta hen hao de wanquan gaixie.

This gives the following hierarchy: (197) laoshi-shuo ‘honestly’ > buxing ‘unfortunately’ > xianran ‘evidently’ > xianzai ‘now’ / yexu ‘perhaps’ > mingzhide ‘wisely’ > yiban ‘usually’ > changchang ‘often’ > yijing ‘already’ > bu-zai ‘no longer’ > zongshi ‘always’ > yizhi ‘continuously’ / ganggang ‘just’ > wanquan ‘completely’ > hao ‘well’

1.7.6 Albanian speech act adverb > evaluative adverb (198) a. Singerisht ka mjerisht njé opinion jo té miré pér ju.!!! sincerely has unfortunately one opinion not good of you ‘To be earnest he unfortunately hasn’t a good opinion of you.’ b. *Mjerisht ka singerisht njé opinion jo té miré pér ju. unfortunately has sincerely one opinion not good of you ‘Unfortunately he to be earnest hasn’t a good opinion of you.’

42 ADVERBS AND FUNCTIONAL HEADS

evaluative adverb > ndoshta ‘perhaps’ (199) a. Fatmirésisht Beni ndoshta do t-ia hedhé edhe kété radhé. fortunately Beni perhaps FUT Subj-cl make it even this time ‘Fortunately B. perhaps will make it even this time.’ b. *Ndoshta Beni fatmirésisht do t-ia hedhé edhe kété radhé. perhaps Beni fortunately FUT Subj-cl make it even this time ‘Perhaps B. fortunately will make it even this time.’

speech—time adverb > ndoshta ‘perhaps’

(200) a. Tani ndoshta Beni ka ikur. “Now perhaps B. has left.’ b. *?Ndoshta tani Beni ka ikur. ‘Perhaps now B. has left.’ habitual adverb > negative adverb (201) a. Né orén 2, Beni zakonisht as nuk ka ngréné ende. at two, B. usually not not has eaten yet ‘At two, B. usually hasn’t eaten yet.’ b. *Né orén 2, Beni as nuk ka ngréné zakonisht ende. ‘At two, B. not not has eaten usually yet.’

ende ‘still’ > gjithnjé ‘always’ (202) a. Ai ende gjithnjé ka fuqi. ‘He still always has power.’ b. *Ai gjithnjé ende ka fuai. ‘He always still has power.’ gjithnjé ‘always’ > térésisht ‘completely’ (203) a. Ai nuk ikupton gjithnjé térésisht vérejtjet. he not them understands always completely remarks ‘He does not always understand the remarks completely.’ b. *Ai nuk 1 kupton térésisht gjithnjé vérejtjet. he not them understands completely always remarks

pjesérisht ‘partially’ > miré ‘well’ (204) a. Ka ribéré pjesérisht gjithcka miré Bent. has redone partially everything well B ‘Beni has partially done everything well again.’ b. *Ka ribéré gjithcka miré pjesérisht Beni. has redone everything well partially B

The overall order thus seems to be: (205) singerisht ‘sincerely’ > mjerisht ‘unfortunately’ > tani ‘now’ > ndoshta ‘perhaps’ > zakonisht ‘usually’ > as ‘not yet’ / ende ‘still’ > gjithnjé ‘always’ > térésisht ‘completely’ > miré ‘well’

1.7.7. Malagasy In Malagasy, a VOS Austronesian language, some (‘lower’) adverbs follow the verb (and its complements) before the subject. The rest precedes the verb, except for speech

ON THE RELATIVE ORDER OF ADVERB PHRASES A3

act ones, which appear at the very end of the sentence. Malagasy is particularly interesting in that the adverbs which precede the verb appear in the same order as in Romance/Germanic while those that follow the verb (and its complements) appear in an order which is the exact opposite of the Romance/Germanic order (Rackowski 1996).

Compare (206), a subset of the adverbs in (142), with (207) (rom Rackowski 1996, 10):

1235678 2348765

(206) speech act > generally > already > no longer > always > completely > well

(207) matetika> efa> mbola> V (QO) > tsara > tanteraka > foana > intsony >

generally already still well completely always anymore j ve speech act

As Rackowski suggests, this systematic reversal of the order of (some) “lower” adverbs can be made sense of if 1t is assumed that the VP raises across the first adverb (tsara ‘well’), the phrase containing V (O) tsara raises across the next adverb up (tanteraka ‘completely’), and so forth, till the adverb intsony ‘anymore’ is crossed over.!!? This situation recalls that of postnominal APs in the Indonesian (Austronesian) DP, and in that of other languages, which appear in exactly the opposite order of the prenominal APs of Germanic; an order which was taken in Cinque (1996, § 4) to arise similarly via successive raisings of increasingly larger XPs across each AP. With APs, the cross-linguistic generalization is that only postnominally APs may appear in the mirror image order of the prenominal ones (cf. Greenberg’s 1963 universal 2Q, and its derivation in Cinque 1996, within Kayne’s 1994 antisymmetric theory). We are led to expect a similar situation with AdvPs.

2

A Case for Adverb Phrases in Spec

Although the adjunction of AdvPs to (nonargumental) maximal projections is commonly assumed in the current literature, there seem to be both conceptual and empirical reasons for locating them in the (unique) Spec positions of distinct maximal

projections. The relative order of AdvPs established in the previous chapter, in interaction with the distribution of (active) past participles and finite verbs in Italian, offer an empirical argument for locating AdvPs in Spec. Before examining this argu-

ment, J mention possible conceptual considerations also favoring the location of AdvPs in Spec. A system that countenances both specifiers and adjuncts is clearly less restrictive than a system that does away with one or the other (while still expressing all the correct empirical generalizations). Suppose we find positive evidence for locating some adjunct XPs mn Spec. Then the desirable possibility arises of doing away with the competitor (adjunction) entirely. See Cinque (1992b, 1994) and Sportiche (1993) for general discussion. One desirable direct formal consequence of Kayne’s (1994) restrictive version of X-bar theory—more generally of his antisymmetric view of syntax—is precisely the availability of a single specifier per projection (or adjunct, their difference being in fact neutralized). Also, while the existence of a rigidly fixed relative order of AdvPs is entirely unexpected under adjunction, it is understandable under the “location-inSpec” hypothesis, at least 1f it can be argued to follow from the fixed relative order of the heads via the general Spec/head agreement relation (as we shall attempt). Moreover, the fact that AdvPs are arguably on left branches is something that would have to be stipulated under the adjunction hypothesis, whereas it follows under 44

A CASE FOR ADVERB PHRASES IN SPEC 45

the “location-in-Spec” hypothesis as Specs are normally (necessarily in Kayne’s theory) taken to be on left branches.! Let us now consider the empirical argument, based on the presence of one head position to the immediate left and one head position to the immediate right of each AdvP.

2.1 Active past participle movement in Italian Restricting our attention, for the time being, to the “lower” pre- VP AdvPs isolated in §1.1, we observe that an active past participle can be found preceding, or following, each AdvP in the sequence (except for the last two, tutto and bene, to which I shall return). Consider (1): (1) a. Daallora, non hanno rimesso di solito mica pil sempre completamente tutto bene in ordine. b. Da allora, non hanno di solito rimesso mica pil sempre completamente tutto bene in ordine. c. Da allora, non hanno di solito mica rimesso pi sempre completamente tutto bene in ordine. d. Da allora, non hanno di solito mica pit: rimesso sempre completamente tutto bene in ordine. e. Da allora, non hanno di solito mica pil sempre rimesso completamente tutto bene in ordine. f. Da allora, non hanno di solito mica pil sempre completamente rimesso tutto bene in ordine. ‘Since then, they haven’t usually not any longer always put everything well in order.’

In the spirit of Pollock’s (1989, 1993) analysis, according to which AdvPs are assumed to occupy fixed positions and what is taken to move is the verb, the word orders

shown in (1) suggest the presence of a distinct head position to the left of the habitual adverb di solito and between all the following adverbs (except tutto and bene). Example (2) suggests the same conclusion relatively to (mica) gia (and piu) (which are exemplified separately, as gid does not fit well, semantically, in (1)). (2) a. Non ha mica gia ricevuto pit niente. b. Non ha mica ricevuto gia pit niente. ‘He has not already any longer received anything.’

In other words, (1) and (2) point to the structure in (3), where the capital X’s indicate head positions: (3) [X[solitamente X [ mica X [ gia X [ pit. X | sempre X [ completamente X [| tutto

bene [vp] ]]JII19) If correct, this conclusion provides the promised evidence for the “AdvP-in-Spec hypothesis,” tn that classical X-bar theory (Chomsky 1970; Kayne 1994) leaves room for just one XP specifier in between two X°’s (heads).* If AdvPs were adjoined to (possibly different) maximal projections, one would not necessarily expect the past participle to be able to appear between (virtually) any two AdvPs.

46 ADVERBS AND FUNCTIONAL HEADS

This conclusion depends on the assumption that verbs move and adverbs stay put—which is by no means a necessary (or universally accepted) assumption. In a moment, I will consider an alternative that rejects this assumption and will compare the different predictions of the two analyses. But first I want to comment on the unexpected behavior of tutto and bene, for, as (4) shows, an (active) past participle cannot be found between them or to the right of bene: (4) a. *Da allora, non hanno di solito mica pit sempre completamente tutto rimesso bene in ordine. b. *Da allora, non hanno di solito mica pit. sempre completamente tutto bene rimesso in ordine. ‘Since then, they haven’t usually not any longer always everything well put in order.’

This fact could be interpreted as suggesting that even under the AdvP-in-Spec hypothesis, one would still need adjunction, as there is no evidence for a head position between tutto and bene or for one to the right of bene, which would create room for two specifier positions hosting them. But there is a more interesting interpretation of (4) that is compatible with tutto and bene each being 1n a distinct specifier position, like all the other AdvPs.That ts, in Italian, (active) past participles must move to the head to the left of tutto (having passed through the head to the nght of bene and between tutto and bene), after which they may raise to higher heads in what looks like optional movement.° Although Italian offers no direct evidence for such lower heads, other Romance varieties do, filling in the links missing in (3). So, for example, in the variety of Logudorese Sardinian spoken in Suni (Patrizia Ruggiu, personal communication), an active past participle can be found to the nght of fottu (tutto) ‘everything’ and to the left of bene, although it cannot occur to the right of bene:* Logudorese Sardinian

(5) a. *Apo bene mandigadu. ‘I have well eaten.’ b. Apo mandigadu bene. ‘I have eaten well.’ c. Apo tottu mandigadu. ‘IT have everything eaten.’ d. Apo mandigadu tottu. ‘I have eaten everything.’

This variety, then, provides evidence for a head position between tutto and bene. French instead offers evidence for a head position to the right of bien (bene) (presumably distinct from the “base” position of the past participle in the VP, as this can raise past at least some other AdvPs (presque ‘almost’, a peine ‘hardly’, souvent ‘often’, etc.), as pointed out in Pollock (1989, 417):

French (6) a. llena bien compris a peine la moitié. ‘He has of it well understood hardly half.’ b. *Ilenacompris bien a peine la moitié. ‘He has of it understood well hardly half.’

As a matter of fact, various Romance varieties appear to differ regarding the particu-

lar head to which the movement of an (active) past participle is obligatory (see Appendix 1).

A CASE FOR ADVERB PHRASES IN SPEC 47

Our analysis so far, in the spirit of Pollock’s (1989, 1993), takes the word order pattern of (3) to be a function of the movement of the past participle over one or more AdvPs, ordered in a rigidly fixed sequence.° Let us now compare this analysis (which JI will call Alternative I) with one that, instead, takes (3) to be caused by the free adjunction of AdvPs to (or by their generation in the multiple Specs of) different maximal projections, to either the left or the right of the past participle, which occupies a fixed position (call it Alternative I). Consider first the following facts. In Italian, an adverb like mica can be found either preceding or following an (active) past participle (see (7a—b)). And the same is true for an adverb like pi (see (8a—b)):

(7) a. Non hanno mica mangiato. ‘They haven’t not eaten.’ b. Non hanno mangiato mica. ‘They haven’t eaten not.’ (8) a. Non hanno pid mangiato. ‘They haven’t any longer eaten.’ b. Non hanno mangiato piu. ‘They haven’t eaten any longer.’ If analogy were at work (if language acquisition were based on analogy), we would expect both of the following sentences, which combine (7a) with (8b) and (7b) with (8a) to be acceptable. But only one is grammatical—the one combining (7a) with (8b):

(9) a. Non hanno mica mangiato pit. “They haven’t not eaten any longer.’ ((7a) + (8b)) b. *Non hanno pit mangiato mica. ‘They haven’t any longer eaten not.’ ((7b) + (8a))

Alternative II (the one with free adjunction, or multiple Specs) offers no immediate solution to the puzzle. If anything, the possibility of freely generating mica and pitt both in position A, to the left of the past participle, and in position B, to the right of the past participle leads us to expect (9b) to be possible. We need to complement Alternative II with an additional filter so that the otherwise free generation of mica and piu to either the left or the right of the past participle would lead to ungrammaticality just in case piz# is generated in position A, on the left, and mica in position B, on the right. This filter, however, simply recapitulates the ordering principle needed to account for the orders possible in position A (see (10)) and in position B (see (11)), where mica precedes piiz:® (10) a. Non hanno mica pit mangiato. ‘They haven’t not any longer eaten.’ b. *Non hanno pit mica mangiato. ‘They haven’t any longer not eaten.’ (11) a. Non hanno mangiato mica pit. ‘They haven’t eaten not any longer,’ b. *Non hanno mangiato pit mica. ‘They haven’t eaten any longer not.’ Alternative I needs nothing special to account for (9b), as it simply offers no way of deriving (9b). The only way for the past participle to precede mica is for it to raise to a head higher than mica. But given that pia is lower than mica in the unique, fixed,

sequence of specifiers, pi could not possibly be found to the left of the past participle.’ Instead, (7), (8), (9a), (10a), and (11a) all are correctly derived by raising the past participle to different heads on the fixed sequence of AdvPs (as noted in note 6, the order of the AdvPs/specifiers does not need to be stipulated under Alternative I if it can be made to follow from the order of the respective heads).

48 ADVERBS AND FUNCTIONAL HEADS

Alternative I is also conceptually superior to Alternative II in that it is inherently more restrictive. It excludes on principled grounds the existence of several possible Romance varieties not excluded by Alternative II, thus exposing itself to being falsified more substantially then Alternative IL. For example, it predicts that no Romance variety could exist, which is (virtually) the inverse of Logudorese Sardinian; compare (5), repeated here as (12), with (13): Logudorese Sardinian (12) a. *Apo bene mandigadu. ‘T have well eaten.’ b. Apo mandigadu bene. ‘I have eaten well.’ c. Apo tottu mandigadu. ‘I have everything eaten.’ d. Apo mandigadu tottu. ‘I have eaten everything.’ Nonexisting variant

(13) a. Apo bene mandigadu. ‘T have well eaten.’ b. *Apo mandigadu bene. ‘I have eaten well.’ c. “Apo tottu mandigadu. ‘I have everything eaten.’ d. Apo mandigadu tottu. ‘I have eaten everything.’ The reason no variety could display the pattern in (13) under Alternative [1s that if the bare quantifier (tout/tutto/tottu/etc.) occupies (comes to occupy) a specifier position higher than the one hosting bene and if the past participle is found to the right of bene (13a), it can occur to the right of the bare quantifier. In other words, under Alternative I, the grammaticality of (13a) implies that of (13c). Note that Alternative II has nothing to say about (13). Under that analysis, the adjunction of a certain AdvP (by generation or movement) to the left or the right of the past participle is not theoretically constrained (apart from the ordering principle mentioned in note 6, not relevant here). In other words, there is nothing under Alternative JI that forces the bare quantifier to be adjoined to the left of the past participle if bene 1s. Hence, as a matter of principle, a pattern like (13) is not excluded under that alternative. For similar reasons, Alternative I, but not Alternative IT, excludes a Romance variety displaying the pattern in (14) (say, another variant of Logudorese Sardinian): Nonexisting variant

(14) a. Apo bene mandigadu. ‘I have well eaten.’ b. *Apo mandigadu bene. ‘I have eaten well.’ c. Apo tottu mandigadu. ‘I have everything eaten.’ d. Apo mandigadu tottu. ‘I have eaten everything.’

If the past participle can appear to the left of the bare quantifier in (14d), it can appear to the left of the lower AdvP bene. In other words, the grammaticality of (14d) implies that of (14b). Again, no comparable prediction is made by Alternative IT. More generally, Alternative I predicts that (1) whenever the order V AdvP; is obligatory, the order V AdvP;,, (where AdvP;, follows AdvP; in the fixed sequence)

is also obligatory; and (2) whenever the order V AdvP; is possible, the order

A CASE FOR ADVERB PHRASES IN SPEC 49

V AdvP;,; also is possible. See Appendix 1 for comparative evidence confirming this prediction. If no Romance variety of the type in (13) and (14) exists, we have one more reason to conclude that the word order patterns in (1) and (2) should indeed be analyzed as required by Alternative I, with movement of the past participle “around” the AdvPs. This, in turn, provides, as noted, a clear argument for locating AdvPs in the (unique) specifier positions of distinct maximal projections.®

2.2 Finite V movement in Italian The preceding argument can be replicated for “higher” AdvPs. Past participles cannot ratse as high, but finite (auxiliary) verbs can. Consider the word order patterns in (15) through (17), parallel to (1) and (2):? (15) a. Miero francamente purtroppo evidentemente formato una pessima opinione di voi. b. Francamente mi ero purtroppo evidentemente formato una pessima opinione di vol. c. Francamente purtroppo mi ero evidentemente formato una pessima opinione di voi. d. Francamente purtroppo evidentemente mi ero formato una pessima opinione di voi. ‘Frankly I unfortunately had clearly formed a very bad opinion of you.’ (16) a. Evidentemente mi ero probabilmente allora formato una pessima opinione di vol. b. Evidentemente probabilmente mi ero allora formato una pessima opintone di voi. c. Evidentemente probabilmente allora mi ero formato una pessima opinione di voi. ‘Clearly I probably had then formed a very bad opinion of you.’

(17) a. Allora aveva forse saggiamente deciso di non presentarsi. b. Allora forse aveva saggiamente deciso di non presentarsi. c. Allora forse saggiamente aveva deciso di non presentarsi. “Then he had perhaps wisely decided not to go.’

Here, the finite auxiliary can be found to the left of the “pragmatic” adverb francamente ‘frankly’, between all of the following “higher” adverbs, and to the right of the lowest of these, the “subject-oriented” saggiamente ‘wisely’. Under Alternative I, this word order pattern, taken to be a function of the movement of the finite V “around” the AdvPs, suggests the structure shown in (18): (18) [xX [francam X [ purtroppo X [ evidentem X [{ probabilm X [ allora X [ forse X

[ saggiam X...}] J] 111] Once again, we can check the superiority of Alternative I to Alternative II, which would freely adjoin these “higher” adverbs (or generate them in multiple Specs) to either the left or the right of the finite V taken to occupy a fixed position. Consider the facts of (19). Given that a “subject-oriented” adverb like saggiamente ‘wisely’ can precede a finite auxiliary (19a) and that an “evaluative” adverb like fortunatamente ‘luckily’ can follow it (19b), we can expect it to be possible to combine the two orders, yielding (19c): (19) a. Gianni saggiamente ha accettato. ‘G. wisely has accepted.’ b. Gianni ha fortunatamente accettato. ‘G. has luckily accepted.’ c. *Gianni saggiamente ha fortunatamente accettato. ‘G. wisely has luckily accepted.’

50 ADVERBS AND FUNCTIONAL HEADS

But (19c) is ungrammatical. If the auxiliary occupied the same position in (19a—b) and “subject-oriented” and “evaluative” adverbs were freely adjoinable to either the left or the right of the auxiliary, this result would be unexpected. Some filtering principle would be needed to reduce the overgeneration induced by free adjunction (or free generation in multiple Specs). The ungrammaticality of (19c) is, instead, expected under Alternative I. In that analysis, (19c) simply cannot be derived. “Evaluative” adverbs are in a Spec higher than that hosting “subject-oriented” adverbs. The position of the auxiliary in (19a~b) only seems the same. In (19a), it occupies a lower head than in (19b). Another use of mica (the preverbal one)—not considered in the previous chapter—may be construed as additional evidence that the finite verb does not occupy a fixed position in the clause. Besides occurring postverbally in construction with a preverbal non (see note 4 of chapter 1), mica can occur in front of a finite verb (in which case it is incompatible with non):'° (20) Gianni (*non) mica (*non) gli telefonera. G. (not) not (not) to-him will-telephone

First, note that the word order of (20) ts not the effect of focalizing the negative element and dislocating the subject, as is arguably the case in (21): (21) Gianni, NIENTE gli ha dato. G., nothing (focus) to-him has given

The reason is that a focalized XP can precede the subject (22a), whereas mica cannot (22b), and that focalized negative phrases can co-occur with a preverbal non (23a), whereas mica cannot, as noted, (23b), and that focalization is “unbounded” (24a), whereas the “movement” of mica is not (24b): (22) a. NIENTE Gianni gli ha dato! nothing (focus) G. to-him has given b. *Mica Gianni gli telefonera. Gli scrivera solo.!! Not G. to-him will-telephone. He will only write to him.

(23) a. NIENTE (non) gli ha dato. nothing (focus) (not) to-him has given b. Mica (*non) gli telefonera. not (not) to-him will-telephone

(24) a. NIENTE ha detto che gli dara. nothing (focus) (he-)said that to-him (he-)will-give b. *Mica ha detto che gli telefonera.!? not (he-)said that to-him (he-)will-telephone

The preverbal position of mica in (20) is also not due to its being “base generated” in (or having moved to) an IP preverbal position. If this were the case, we could expect (26) to be possible alongside (25):

(25) Gianni non l’ha fortunatamente mica trovato. ‘G. hasn’t luckily not found it.’ (26) *Gianni mica |’ha fortunatamente trovato. ‘G. not has luckily found it.’

A CASE FOR ADVERB PHRASES IN SPEC 5]

Of course, under the “base generation” option, (25) could be excluded by recourse to the independent semantic filter mentioned in note 5 and under the movement option by Relativized Minimality.! The point, however, is that Alternative I accommodates the ungrammaticality of (26) more naturally and directly, by simply assuming that verbs, not AdvPs, can occupy different positions within a certain (here, the post-C° and pre-VP) “space.” This means that mica has not really moved at all. It is the verb that has stopped in the head corresponding to the Spec occupied by mica, as witnessed by the fact that this preverbal mica is preceded and followed by exactly the same adverbial classes that precede and follow the postverbal mica (compare (1) and (15-17) with (27)):"4 (27) a. Gianni purtroppo forse stupidamente mica gli ha pit telefonato. ‘G. unfortunately perhaps stupidly not to-him has any longer telephoned.’ b. *Gianni purtroppo forse mica gli ha stupidamente pit telefonato. ‘G. unfortunately perhaps not to-him has stupidly any longer telephoned.’ c. *Gianni purtroppo mica gli ha forse stupidamente pit telefonato. ‘G. unfortunately not to-him has perhaps stupidly any longer telephoned.’

If we are to conclude, then, that AdvPs are indeed in the unique Spec positions of different maximal projections, of what heads are these maximal projections projections? I take up this question in chapter 4, suggesting that they are the functional heads of the clause. This requires establishing the order of clausal functional heads on independent grounds, a task which I take up in chapter 3.

3

On the Order of Clausal Functional Heads

This chapter tries to determine the order of clausal functional heads independently of any considerations involving AdvPs and their relative order. I look at different types of evidence bearing on this question—namely, the order of “nonclosing” (agglutinating) suffixes, the order of “closing” (inflectional) suffixes and auxiliaries, the order of functional particles, and the order of various combinations of these elements in mixed cases.' The picture that emerges is the following. If we ignore agreement and negation, the partial relative orders of functional heads for which there is overt evidence (of one of the above four kinds) in different languages appear to be compatible with a single overall order. Thus, putting the partial relative orders of functional heads found in different languages into a single, more

comprehensive, order seems to provide no contradiction, at least in a more careful examination. I take this to be significant, that is, nonaccidental. In the interpretation J suggest, this is so because the partial orders found overtly in different languages are subsequences of a single universal sequence of functional heads, present in all languages. As this conclusion is far from being obvious (or necessarily correct), I return in the next chapters to some considerations that may support it. The “generative” and “functional’/“typological” literature on the order of suffixes is quite rich, and some of it explicitly uses the order of suffixes to motivate a particular ordering of functional heads, under (a generalized version of) Baker’s Mirror Principle (see Pollock 1989, 1993; Belletti 1990; Chomsky 1995, chap. 2; and much subsequent work of the early 1990s). In §3.1, §3.2 (and Appendix 2, which gives a synopsis of the orders of overt functional heads in selected languages), I review some of this work, with no pre52

ON THE ORDER OF CLAUSAL FUNCTIONAL HEADS 53

tense to exhaustiveness or detail, merely giving some illustrative examples on which

conclusions about the order of functional heads have been, or can be, based, and referring to the cited works for fuller discussion. The only goal of this survey is to see whether the different orders of functional heads motivated for each particular language by its suffix order, under the Mirror Principle, are compatible with one another, thus lending support, under a certain view, to the postulation of a single overall order valid for all languages. In §3.3, I turn to the evidence from free functional morphemes (particles), and in §3.4 to the evidence from mixed cases. We will see that the orders suggested by such cases are compatible with that based on the order of suffixes. In §3.5 I discuss, somewhat speculatively, the status of prefixes vis-a-vis the (generalized) Mirror Principle. A general discussion, and a specific overall order of functional heads, compatible with the different orders motivated separately for each language, are presented in §3.6. In this chapter, I deliberately ignore agreement and negation. They appear to occur in more than one position (sometimes simultaneously), and are thus a potential source of apparent counterexamples to a fixed hierarchy of heads. Their case will be taken up in chapter 5.

3.1. Evidence from the order of “nonclosing” (agglutinating) suffixes

In this section, I consider the case of “nonclosing” suffixes (namely, suffixes that can be stacked one on to the other), beginning with Korean. In the complex verbal form in (1), as described by Sohn (1994, §2.1.3), -hiis a suffix marking passive voice, -si a subject honorific suffix, -ess a suffix marking either anterior or past tense (the pluperfect, or anterior of the past, when duplicated), -keyss a suffix marking “conjectural” (epistemic) modality, -sup an addressee honorific suffix, and -ti an “evidential” mood suffix (see n. 13) used to recall “a fact that one has witnessed” (p. 342) (Sohn calls it “retrospective mood”). Finally, the suffix -kka marks the speech act mood “interrogative” (contrasting with “declarative” -(#)a, and other speech act moods):° (1) Ku pwun-i caphi-si-ess-ess-keyss-sup-ti-kka? the person-NOM catch-PASS-AGR-ANT-PAST-EPISTEM-AGR-EVID-Q ‘Did you feel that he had been caught?’

As shown by (2) and (3), also from Sohn (1994, 354), another class of suffixes may intervene between evidential and speech act mood suffixes (which Sohn calls “apperceptive mood” suffixes). As they carry a sense of “surprise” (pp. 353ff), thus introducing an evaluation of the proposition by the speaker, I will call them “evaluative mood” suffixes: (2) Ku say-ka cwuk-ess-keyss-kwun-a! that bird-NOM die-ANT-EPISTEM-EVALUAT-DECL ‘That bird must have died!’

54 ADVERBS AND FUNCTIONAL HEADS

(3) Minca-nun ttena-ss-te-kwun-yo! M.-TOP leave-PAST-EVID-EVALUAT-POLITE ‘T noticed that M. had left!’

If we ignore the honorific suffixes -si- and -sup- in (1), and the “politeness” suffix -yo in (3), which are plausibly agreements (see the “addressee agreement” of Basque; Oyarcabal 1993), the relative order of the Korean suffixes in (1)—(3), under the Mirror Principle, provides evidence for the order of the functional heads shown in (4): (4) Moodgpecch act > Moodevatuatve > MOOd.ydentias > Modality > T(Past) > T(Anterior) >

Voice (>V)

Turkish is another language with rich agglutinating morphology. The examples shown in (5) point to the order of suffixes in (6), which, again, under the Mirror Principle, suggests the relative order of functional heads shown in (7): (5) a. Ali gel-me-di-mi? (Tura 1981, 317) A. come-neg-PAST-Q ‘Hasn’t A. come?’ b. Oku-y-abil-ecek-ti-m. (Jaklin Kornfilt, personal communication) read-y-MOD-FUT-PAST-|sg ‘I was going to be able to read / I would be able to read.’ c. Oku-n-uyor. (Jaklin Kornfilt, personal communication) read-PASS-PROG ‘It is being read.’

(6) V-PASS-PROG/MOD-FUT-PAST-Q* (7) Mo0dgpecch act > T(Past) > T(Future) > Modality / Aspectyrooressive > Woice > V°

The Turkish evidence for Modality “lower” than Tense seems, at first, to contrast sharply with the evidence from Korean for Modality “higher” than Tense (see (1)). The contrast, however, is only apparent. The modals in the Korean and Turkish sentences (1) and (Sb) are two different types of Modals. The former 1s epistemic, the latter a root modal. Although this traditional semantic difference 1s generally not related to a difference in structure, some evidence exists that the two modals do not occupy the same position and that epistemic modals are higher than root modals. In double modal varieties of English, like Scots English (Brown 1992, 76ff), and in various American varieties (see Boertien 1986; Battistella 1991; Cunningham 1992, 46ff; and Thrainsson and Vikner 1994, among others), the first modal has an epistemic, and the second a root sense:® (8) a. He should can do it (= he likely can do it). (from Brown 1992, 77) b. He must can do it (= he surely can do it).

Confirming evidence comes from the richly agglutinating language Una, an Irian Jaya language of New Guinea. According to Louwerse (1988), the “probabilitive mode”

suffix is outside absolute tense suffixes, which in turn are outside “aptative” or “abilitative” modal suffixes. See (9):’

ON THE ORDER OF CLAUSAL FUNCTIONAL HEADS 55

(9) a. Er bin-kwan-de-darib. (Louwerse 1988, 55) she go-FUT-3sg-PROBAB ‘She might go.’ b. Ni buk-ti-nyi. (Louwerse 1988, 25) I sit-ABIL-PRES ‘T can sit.’

Again, under the Mirror Principle, this suggests the order of heads in (10): (10) Modepistemic > T(absolute) > Modyoot *

Chinese, though far from being a richly agglutinating language, nonetheless displays some cases of “stacked” suffixes, such as the “completive” and “perfect” suffixes shown in (11) (from Smith 1987, 96ff):? (11) Ta zuotian xie-wan-le yifeng xin. he yesterday write-COMPL-PERF a letter ‘He wrote a letter (to the end).’

The Korean, Turkish, Una, Tauya, and Chinese examples presented here are thus compatible with the single overall order shown in (12): (12) Mood speech act > Moody atuative > Mood. yigentiat > Modgpistemic > T(Past) > T(Future) > Mod, / T(Anterior) > AspeCtpertect > ASPCCtprogressive / ASPECtcomplenve > Voice

(> vyie

In support of this first conclusion, we may add Bybee’s (1985) typological observation that the order of suffixes appears to be consistent across languages (p. 201), with aspect markers closer to the stem than tense and mood markers (p. 34).!!

As to the relative order of tense and mood, Bybee says (p. 35) that, in her sample, tense is generally closer to the stem than mood, with some exceptions. Under mood, however, she collapses “speech act” mood (imperative, interrogative, optative, etc.), evidentials, root and epistemic modality, and “grammatical mood” (subjunctive versus indicative): a heterogeneous class of elements, which, in fact, appear to occupy distinct positions (some lower, and some higher, than Tense). This, I would argue, is at the basis of the exceptions she alludes to, and of the apparent inconsistency she mentions between her generalization on suffix ordering and Foley and Van Valin’s system (1984). Once finer distinctions are introduced, and the terminology is clarified, Bybee’s and Foley and Van Valin’s generalizations appear to become consistent. The elements falling under Mood in Bybee’ s system are distributed in Foley and Van Valin’s (1984, 213ff) into three grammatical categories: [llocutionary Force (corrresponding to what I called above “speech act” mood), Status (essentially expressing realis/irrealis distinctions such as indicative versus subjunctive), and (root) Modality (mental and physical ability, permission, etc.).!* They claim that Modality is closer to the verb stem than either Status or Tense (p. 215f). This, they claim, is suggested for the relation between Modality and Tense in such languages as Lisu (Sino-Tibetan), among others (note that (13) involves root modals): (13) a. Asa nya ami khwa kwa-a. A. TOP field hoe mentally able-nonpast ‘A. is able (knows how) to hoe the field.’

56 ADVERBS AND FUNCTIONAL HEADS

b. Asa nya ami khwa da-a. A. TOP field hoe able-nonpast ‘A. is (physically) able to hoe the field.’

As to the relative ordering between Status (grammatical mood) and Tense, Foley and

Van Valin tentatively suggest (1984, 216ff) that the former is mside the latter (a suggestion that will be supported in the next sections). They then consider the category of evidential mood suffixes (which, when present, appear after Tense suffixes— see (14)-(17)),' and Ilocutionary Force suffixes, which are the outermost suffixes, only followed by subordination suffixes (if any)—-see n. 3 and Foley and Van Valin (1984, 220ff).!4 (14) a. fra-a-n. (Kewa (Indo-Pacific), from Foley and Van Valin 1984, 218) cook-3sgPAST-seen ‘He cooked it (I saw it).’ b. fra-a-ya. cook-3sgPAST-unseen ‘He cooked it (hearsay, I didn’t see it).’ (15) a. John Tiirkiye-ye gid-ecek-mis. (Turkish, from Yavas 1980, 41) J. Turkey-to go-FUT-EVID ‘Reportedly, John will go to Turkey.’ b. Yagmur ya$-acak-mis. (Turkish, from Aksu-Kog and Slobin 1986, 161) rain[NOUN] rain[ VERB]-FUT-quotative evid ‘It is reported that it will rain.’ (16) Kaya shamu-nga-shi. (Ecuadorian Quechua, from Muysken 1977, 27) tomorrow come-3FUT-hearsay “They say he’ ll come tomorrow.’

(17) Tukua-tuka-na-puga-vaaci. (Ute (Uto-Aztecan), from Givén 1982a, 40) meat-eat-HAB-PAST-EVID (hearsay) ‘(She) used to eat meat (so I hear).’

To summarize, the data seen so far, and Bybee’s (1985) and Foley and Van Valin’s (1984) typological generalizations, are compatible with the overall relative order of verbal suffixes shown in (18), which gives evidence, under the Mirror Principle, for the relative order of functional heads shown in (19): (18) V-voice / compietive aspect / progressive aspect—perfect aspect—anterior tense / habitual aspect / root modality-grammatical mood—future tense—past tense—epistemic modality—evidential mood—evaluative mood-—speech act mood!» (19) Mood, peech act > Moodgyatuative > Moody dential > Modenistemic > T(Past) > T(Future) > Mood wreahss > Mod,oot / ASPeCthapimnal / T(Anterior) > ASP€Cthertect > ASPECtopressive

/ Aspectyompietive / Voice > V

We now turn to the case of “closing” suffixes and auxiliaries, to see whether the evidence for the relative order of functional heads based on them 1s compatible with (19).

ON THE ORDER OF CLAUSAL FUNCTIONAL HEADS 57

3.2 Evidence from the order of “closing” (inflectional) suffixes and auxiliaries In inflectional (as opposed to agglutinating) languages, suffixes typically close off the word (or require, as in Romance, a word-marker to close off the word; see Harris 1991), disallowing further affixation. As a consequence, when more suffixes occur in a sentence, more verbs are needed to “bear” them (typically, one for each functional suffix).!® In “head-initial’” languages like English and Spanish, where no (successive) left-

ward movements of lower portions of the clause plausibly take place (Kayne 1994), the sequence of the various V + (functional) suffix combinations may thus be taken

to provide direct evidence for the relative order of the corresponding functional heads.!’ If so, a sentence such as (20a) in English, and its equivalent (20b) in Spanish, give evidence for the order of functional heads shown in (21):!8 (20) a. These books have been being read all year. b. Esos libros han estado siendo leidos todo el afio. (21) Tense > Aspectpertect > ASPCCtprogressive > Voice (> V)"”

We expect to find the mirror-image situation in those “head-final” languages in which suffixes also “close off” the word (requiring insertion of auxiliaries), and in which leftward movements of nonheads apply successively, as shown in (22): a derivation which Kayne (1994, §5.5) originally suggested as the other possible source (in addition to the more familiar head-raising type) of “sequences of inflectional morphemes in a “head-final” language” (p. 52):

(22) ...X[yp...YZP]...3...X[ypZPYt]...3...[ypZP Y typ ] X typ Here I take the nonhead-raising derivation in (22) to be operative in “head-final” languages like Hindi, where sequences of V + functional suffix combinations are found which are the mirror image of the English case seen in (20a). See (23) and (24): (23) a. Kis-ko raam-ne socaa ki siitaa-ne dekhaa thaa. (Mahajan 1990, 39) who Raam thought that Sita see-ANT be.PAST “Who did R. think that S. had seen?’ b. Raam roTu khaataa rahtaa thaa. (Mahajan 1990, 78) Raam bread eat PROG be.PAST *R. used to keep on eating bread.’

These cases seem best analyzed as involving verbal heads (Aux + suffix) which force their complements to move leftward to a specifier in between them and the next higher head. As Kayne notes (1994, 53), the type of derivation in (22) “produces a YX that is not a constituent,” and, indeed, the sentence-final sequence of verbs in Hindi does not form a constituent, as Mahajan (1989, 225ff) explicitly notes. The verbs can, for

example, be separated by various types of lexical material (“the negation word, emphatics and some adverbs’), and each can be moved to the beginning of the sentence separately from the others.

58 ADVERBS AND FUNCTIONAL HEADS

In other “head-final” languages, like standard German, the sentence-final verbs cannot be separated by any material. This could be due to the incorporation of each verbal head to the left of the next higher one, to eventually form a constituent, with the non-head subparts of IP raising leftward over the verb cluster, either individually (Kayne 1994, 141, n. 15; see also Zwart 1993, 334 ff), or within a single constituent. The resulting order would also be the mirror image of that found in English, due to the successive incorporations of the verbs to the left of the next one up: (24) ...dass er von der Bank angestellt worden sein muss. ... that he by the bank employed been have must ‘,.. that he must have been employed by the bank.’

Alternatively, the strict adjacency of the sentence-final verbs could be due to the verb forcing its complement to raise to its Spec (as Kayne 1994 suggests could be the case in agglutinating languages). This derivation is compatible with the idea that AdvPs fill the (unique) Spec of functional projections if the “auxiliary” verbs are generated ina VP subjacent to the functional head to which they ultimately raise at LF, and the complement of the auxiliary V raises to the Spec of such VP.?! The difference between German-type languages and Hindi-type languages would then reside in the Spec position to which the complement of the verb raises (whether or not it is the one adjacent to the V). I take agglutinating languages like Korean, Japanese, and Turkish to be different from both German and Hindi in that they involve successive head raisings of bound

forms to build up a word (followed by the raising of the word’s complement to a higher Spec). When some free morpheme happens to fill a particular functional head, thus interrupting the successive raising of a lower bound morpheme, the bound form is “closed off,” and any bound forms corresponding to heads higher than the free morpheme will require the insertion of an auxiliary. See the case of Korean discussed in n. 8, and that of Turkish discussed in §3.4, and Kang’s (1988, chap. 2) notion of “morphological closure.” If so, “head-final” languages provide the same evidence as English (and Spanish) for the order of functional heads, as shown in (21).

3.3. Evidence from the order of functional particles Free functional morphemes (particles), in contrast to bound functional morphemes, bar adjunction of the immediately lower head (but see n. 27 for an apparent case of a particle adjoining to another) and also prevent it from raising past them.?? This means

that in “head-initial” languages, particles allow us to directly observe the order of functional heads. One such case is provided by creole languages, which characteristically display all their functional particles before the verb.*° In the literature on Creoles, it has long been observed, and assumed, that the ordering of such particles is T(ense)-M(ood/modal)—A (spect) (see Bickerton 1974, and Singler 1990 for even earlier sources of the same observation; see also the articles contained in Singler 1990). This ordering is essentially endorsed by Muysken (1981), who, in fact, suggests, after Woisetschlaeger (1977), that a principle of Universal Grammar requires that Aspect be interpreted before Mood, and Mood before

ON THE ORDER OF CLAUSAL FUNCTIONAL HEADS 59

Tense. (See the similar suggestions on verbal suffixes by Foley and Van Valin, and Dik et al. mentioned in the preceding nn. 11 and 14.) Though in essence correct, this ordering is somewhat gross and must be qualified. For one thing, various aspectual particles can co-occur, so that their relative order needs to be determined. For another, there are occasional claims in the litera-

ture for the order Modal > Tense rather than Tense > Modal (see, for example, Gibson’s 1986 study of Guyanese Creole). I begin with Guyanese Creole, which is particularly rich in functional particles and which Gibson (1986) takes, as noted, to instantiate a Modal > Tense > Aspect order (in contrast to other creoles).

3.3.1 Guyanese Creole The basilect has four distinct aspectual particles: a for progressive aspect, de for durative aspect, a for habitual aspect, and don for what is probably better termed “relative or anterior tense” (in the sense of Reichenbach 1947). See the discussion in chapter 4, §4.15).74 The two a particles must be kept distinct. They can co-occur, separated by the

durative aspect particle de, and by such adverbs as aalweez ‘always’ and neva

‘never’:

(25) Shi a aalweez/neva de a sing. (Gibson 1986, 582ff) she HAB always/never DUR PROG sing ‘She usually always/never keeps singing.’

The particle don precedes the particle de, though its order with respect to the habitual particle a is not discussed by Gibson:*° (26) Da taim dem don de somwee a big maarkit. ((a) of Gibson 1986, n. 7) ‘By that time, they are already somewhere in Big market.’

This gives us the partial order in (27): (27) ... Aspectyapua/ TCAnterior) > Aspectauranve > ASPCClprogressive > V

Aspectual particles follow (absolute) Tense particles. In Guyanese, there are two such particles—gu, for the future, and bin, for the past—which in fact turn out to combine, to yield a future of the past or conditional meaning (with the order bin gu). See Gibson (1986, 574):2/ (28) Jaan bin gu rid. (Gibson 1986, 585) John PAST FUT read ‘J. would have read.’

Finally, Gibson mentions a class of modals (‘of condition’’), shud(a) ‘should (have)’ ,

mait(a) ‘might (have)’, kud(a) ‘could (have)’, which are found preceding Tense (e.g., bin ‘Past’, though it is possible that it is the lower Anterior; see chapter 4 for discussion): (29) Jaan shuda bin kyaan get fu gu. (Gibson 1986, 585) J. MODepisteme PAST MOD, MOD, go

‘J. should not have been able to be allowed to go.’

60 ADVERBS AND FUNCTIONAL HEADS

They appear to be epistemic, thus patterning as the Korean epistemic modal suffixes, which were seen above to be outside tense suffixes.”* Putting these various elements together, we arrive at the following order of functional heads for Guyanese Creole: (30) Modaenstemic > T(Past) > T(Future) > Asppapitnal / T(Anterior) > ASpguratve >

(shud) (bin) (gu) (a) (don) (de)

ASP progressive >V

(a)

3.3.2 Sranan The neighboring creole Sranan gives us apparently explicit evidence that root modals occupy a position to the right of the T(Past) and T(Future) particles. See (31a—b), from Seuren (1983, 227), and (32), from Winford (1998b):

(31) a. A ben kan nyan. he PAST can eat ‘He could eat.’ b. A ben o kan nyan. he PAST FUT can eat “He would be able to eat.’

(32) a. Yu ben musu gwe a gron nanga boto... ( = (3)a of Winford 1998b) you PAST must go LOC ground with boat “You had to go to the [planting] ground by boat... Modals, however, can also occur to the left of the (Past) Tense particle ben (see (33a), from Seuren 1983, 227), and (33b) from Winford (1998b), in which case they appear to be interpreted epistemically (but see Winford 1998b for a fuller picture):

(33) a. A kan bene nyan. he may PAST PROG eat ‘He may have been eating.’ b. Kande a musu ben Iasi, yu si. ( = (8)b of Winford 1998b) perhaps it must PAST lose, you see ‘Maybe it must have got lost.’

This might be taken as evidence for an order Modgpistemic > T(Past) > T(Future) > Mod,oot / ASPprogressive > V, reminiscent of the situation found in Korean. But this conclusion is not necessarily right. In most creoles, a verb without particles is typically understood as non-past if stative and as past if nonstative (see Bickerton 1974). This appears true also for Sranan, for which Seuren renders a form like mi nyan (lit.: ‘T eat’) as ‘I have eaten /T ate’, and mi ben nyan as ‘I had eaten’ (the present being mi e nyan ‘lam eating’). See also Haitian Creole in the following section. Seuren (1983, 227ff and n. 8) analyzes a sentence like (34) as containing underlyingly two occurrences of ben—one marking “past” (ben,) and the other “past perfect” (ben,)—the second of which is deleted by rule: (34) A ben sa nyan. (from A ben; sa ben,—>0O nyan) he PAST FUT eat ‘He would have eaten.’

ON THE ORDER OF CLAUSAL FUNCTIONAL HEADS 61

Evidence for that comes from such sentences as (35a), where ben surfaces to the right of sa, and more clearly from such (rarer) cases as (35b), from Adamson and Smith (1995, 229), where both ben’s are simultaneously realized (see also the case reported in Seuren 1981, 1054):

(35) a. Asa ben nyan. (see (34)) he FUT PAST (Anterior) eaten ‘He will have eaten.’ b. A ben o ben e dray. he PAST FUT ANT PROG turn ‘He would have been turning.’

Generalizing Seuren’s ben deletion rule, it may be possible to account for the ambiguity of mi nyan by taking it to derive from either mi ben; > O nyan ‘[ ate’ or mi ben, — 0 nyan ‘T have eaten’.”’ If this is so, the possibility arises that (33) contains an instance of ben, (plausibly to be identified with T(Anterior)) rather than with ben, (T(Past)); in which case, no argument can be made for a Mody stemic preceding Tense (T(Past)). I leave the question open here returning to it after introducing the case of Haitian.

3.3.3 Haitian Creole Hattian Creole also has a rich number of functional preverbal particles for Tense, Mood, and Aspect (see Sylvain 1979; the articles in Lefebvre, Magloire-Holly, and Piou 1982; Leblanc 1989; Spears 1989, 1990; DeGraff 1993, 1994; and Lefebvre 1995, 1996, among others). Aspectual particles are ap, marking progressive aspect, konn, marking habitual aspect, fék (sot) (“venir de,” “to have just”), marking so-called retrospective aspect (see Binnick 1991, 382; Frawley 1992, 322).°° Progressive ap (henceforth ap.) must be distinguished from a homophonous particle marking Future (henceforth, ap,). The two occupy different positions 1n the clause. While ap, (the progressive particle) has to follow such adverbs as deja “already’, poko ‘not yet’, toujou ‘always’, pa janm ‘never’, and others, ap, (the Future marker) has to precede them. See Leblanc (1989, 41, 45), from which (36a) and (36b) are drawn, and Lefebvre (1995, 171; 1996, §1.1):*! (36) a. Jan toujou ap travay fo. J. always PROG work hard ‘J. 1s always working hard.’ b. Mary ap toujou travay. M. FUT always work *‘M. will always work.’

In fact, the two ap’s can co-occur (separated by the mentioned adverb classes).*? See Leblanc (1989, 53, 56), from which (37a—b) are drawn: (37) a. Jan ap souvan ap rakonte yon istwa. J. FUT often PROG tells a story ‘J. will often be telling a story.’

62 ADVERBS AND FUNCTIONAL HEADS

b. Mari p’ap toujou ap travay. M. NEG FUT always PROG work ‘M. will not always be working.’

Summarizing, the partial order of the particles so far examined is: (38) ap,>konn > féek (sot) > ap, Namely, the order of heads: T(Future) > Asppabitual > ASPretrospectve > ASP progressive.

As was the case in Guyanese and Sranan, in Haitian, too, (Future) particles can follow the Past tense particle (te), to yield a future of the past, or a conditional.

See (39a—b), containing the realis future particle ap, and the irrealis one, a/va, respectively:*? (39) a. M pat ap aksepte sa. (adapted from Forrest 1990, 21) Isg NEG PAST FUT accept that ‘I would not accept that.’ b. Mari t’a travay. (= (14b) of Leblanc 1989) M. PAST FUT work *M. would have worked.’

Modals appear to follow the Past and Future Tense marker te and ap, respectively. See (40a—b) and (41a—b):

(40) a. ZA te ka vini. (= (14a) of Magloire-Holly 1982) Jean PAST can come ‘J. was able to come.’ b. Jan te dwe rakonte yon istwa. (= (25a) of Leblanc 1989) J. PAST must tell a story ‘J. was obliged to tell a story.’ (41) a. ZA ap ka vini. (= 14b) of Magloire-Holly 1982)*4 Jean FUT can come ‘J. will be able to come.’ b. Jan ap dwe vini. ( = (30d) of Leblanc 1989) Jean FUT must come ‘Jean will have to come.’

There are some interesting complications, however, which bring us back to the order between Tense and epistemic Modality. The past tense particle te, which was seen to precede dwe in (40b), can also be found following it. See the minimal pair (42a—b) (= (23b-c) of Magloire-Holly 1982, 107, who notes that “te dwe a toujours le sens déontique et dwe te, toujours le sens épistémique’”’):* (42) a. ZA te dwe maize. J. PAST must eat ‘J . has had to (obligation) eat.’ b. Za dwe te mazZe. J. must PAST eat ‘J. must (probability) have eaten.’

It would seem, then, that epistemic modals precede T(Past) while root modals follow it. This conclusion, however, would not follow if modals in Haitian were not

ON THE ORDER OF CLAUSAL FUNCTIONAL HEADS 63

functional particles but main verbs taking a complement clause, as argued for in Magloire-Holly (1982). Possible evidence for that may be a sentence like (43), which Leblanc (1989, 51) reports as being ambiguous between a root reading (43a) and an

epistemic reading (43b), even in the presence of a T(Past) particle preceding the modal:*°

(43) Jan te dwe te rakonte yon istwa. a. ‘Jean était obligé avoir raconté une histoire.’ b. ‘Jean était supposé avoir raconté une histotre.’

Moreover, as already noted earlier for Sranan, no such conclusion would follow either if two te particles existed in Haitian (analogously to Sranan’s two ben’s), one marking T(Past), the other T(Anterior), with one or the other generally deleted. Indeed, some evidence exists for postulating two te’s. So, for example, while fe, (T(Past)) precedes T(Future) (see (39a—b)), and such adverbs as deja (see (44a)), te, (T(Anterior)) follows both. See (44b); see also the different positioning and interpretation of fe with respect to pou discussed in Lefebvre (1996, $1.2): (44) a. Pyé te deja ap domi. (= (5a) of Leblanc 1989) P. PAST already PROG sleep ‘P. was already sleeping.’ b. Mari ap deja te fin domi. (= (40d) of Leblanc 1989) M. FUT already ANT finish sleep *M. will already have finished sleeping.’

In certain cases, they even appear to co-occur, as in (45) (though not all speakers, apparently, accept this possibility; L. H. Desouvrey and M. DeGraff rejected it):?’ (45) Jan te toujou te ap rakonte yon istwa. (= (27a) of Leblanc 1989) J. PAST always ANT PROG tell a story ‘J. had always been telling a story.’

In conclusion, in contrast to Guyanese Creole, neither Haitian nor Sranan may be taken to provide direct evidence for the order Mod, pistemic > T > Mod,oo, (though nei-

ther provides evidence against it, either). To summarize, the different creoles examined above give evidence for the following sequence of functional heads. Left open here is the precise location of root modality with respect to various aspectual heads. (46) Mod.pistemic > T(Past) > T(Future) > Asppabitua /T(Anterior) > Aspetrospective > ASPdurauve > ASP progressive (>V)

This sequence is compatible with the sequence arrived at on the basis of closing and nonclosing suffixes in the preceding sections.

3.3.4 Gungbe Another group of languages displaying preverbal tense and aspect particles are the West African Gbe languages, among which are Fongbe and Gungbe.** In Gungbe, for example, one finds the following preverbal particles:

64 ADVERBS AND FUNCTIONAL HEADS

na This signals the “immiunence of an action’”—“to be about to”—and has to be preceded by the progressive particle t0.°° (47) Asfbé td na x3 méto 15. A. PROG PROSP buy car the ‘A. 1s about to buy the car.’

to (marks progressive aspect) (48) Séna to kiklo moto 15. S. PROG wash car the ‘S. 1S washing the car.’

nd (marks habitual aspect) (49) Séna nd klo méto 15. S. HAB wash car the ‘S. habitually washes the car.’

nd (marks the future tense) (50) Séna n4wahwégbe. S. FUT come home ‘S. will come home.’

ni (marks injunctive/subjunctive and conditional)” (51) a. Séna ni klo méto 15. S. INJ wash car the ‘S. better wash the car!’ b. Un jro na Kofi ni wa. I want for K. SUBJ come ‘I want for K. to come.’ c. Ni Kofi wa Asiba na ton. COND K. come A. FUT go out ‘Tf K. comes, A. will go out.’

The combinations of these particles mentioned in Aboh (1993) area n3 (52a); na to (52b); nd to (52c); na nd té (52d); to na (47); na te na (52e) (see Aboh 1993, 7, 11, 21, 32). Aithough nf and nd cannot occur together, there is evidence that the position occupied by ni precedes that occupied by nd, for ni precedes the negative particle md while nd follows it (see (51f-g). These various partial orders give the consistent overall order (53), corresponding to the order of heads shown in (54), which is compatible with the orders of heads seen so far. (52) a. Asibé nd nd wa hwégbe. A. FUT HAB come home ‘A. will frequently come home.’ b. Séna na to dudu les: [5. S. FUT PROG eat rice the *S. will be eating the rice.’

ON THE ORDER OF CLAUSAL FUNCTIONAL HEADS 65

c. Asibé nd t6 zizé vi le. A. HAB PROG take children the A. is frequently taking the children.’ d. Asfbaé né n5 td kpikpon vi Ié go. A. FUT HAB PROG take care of the children A. will frequently be taking care of the children.’ e. Asiba nd to na x3 kéké 15. A. FUT PROG PROSP buy the vélo ‘A. will be about to buy the vélo.’ f. Kofi ds ni-e d3 ni ma/*ma ni wa blo. K. tell to-him COMP SUBJ NEG / *NEG SUBJ come NEG *K. told him not to come.’ g. Kofi ma na/ *na4 ma wa. K. NEG FUT / *FUT NEG come *K. will not come.’

(53) ni(> maé)>na>nd>td>na(> V) (54) Mood speech act (> NEG) > T(Future) > ASPhabitual > ASP progressive > ASP prospective ( > V)

“Head-final” languages with functional particles should display all the particles sentence finally, in an order which is the mirror image of the one found in comparable “head-initial” languages. This case would be derived (with particles instead of inflected auxiliaries) by the very same process discussed earlier, which derives Hindi sentence-final verb clusters: namely, via successive leftward movements of nonheads across their heads, as shown in (22), from Kayne (1994), repeated here as (55):

(55) ...X[yp..-YZP]...3...X[ypZPYt]......[ ypZP Ytyp ] Xtyp... Some “head-final” languages of the Tibeto-Burman area seem to be of this kind. In Kachin, “the invariable base is followed by grammatical markers which generate distinctions of aspect, mood and tense” (Campbell 1991, 691). Indeed, to judge from the example in (56), the relative order of the Tense particle and of the root Modal particle is the mirror image of that found in the creole languages already discussed (see (41b)): (56) Ngai galaw lu na. Ido MOD FUT ‘I shall have to do this.’

In Mizo, another “head-final” language of the same group, the order of sentencefinal particles for (nonroot) modality and “prospective” aspect, and yes-no questions and T(Future), respectively, is the reverse of that of “head-initial” languages: (57) a. A kall doonl lou ang2. (Chhangte 1986, 141) (s)he go PROSP NEG MOD ‘(S)he may not be going to go.’ b. Coo3 i-n eil angl em2. (Chhangte 1989, 162) food 2NOM-PL eat MOD Q “Will you be eating?’

Similarly, in the Australian language Maranungku, the particle expressing future tense

follows the particle expressing progressive aspect (cf. (58a), and in Basque (which

66 ADVERBS AND FUNCTIONAL HEADS

also shows inflection) the evidential particle omen follows the root modal particle ahal (cf. (58b): (58) a. Yer ngeti tyapat me tu. (Tryon 1970, 46) tomorrow I sit swim PROG FUT “Tomorrow I shall be swimming. b. Zorrak ordaindu ahal omen dituzte. (Ortiz de Urbina 1989, 129) debts pay-off be-able QUOT Aux ‘(It is said that) they can pay off their debts.’

Finally, in Sanio-Hiowe, a “head-final” language of New Guinea, the order of the perfect aspect particle, of the tense particles, and of the interrogative particle appears to be the mirror image of that found in “head-initial” languages with particles. See (59a-b); see also the case of the north American language Hidatsa in Appendix 2: (59) a. Eiworo urei yei awere. (Lewis 1972, 15) who stay FUT Q ‘Who is staying?’ b. Onu tei ye. (Lewis 1972, 17) sit PERF PAST “We sat down.’

3.4 Evidence from mixed cases In this section, some cases will be considered in which particles or auxiliaries co-occur with bound functional morphemes. Given the Mirror Principle and the universal Spec—head—complement order, coupled with leftward movements of heads

(and nonheads in the derivation of “head-final” languages; see Kayne 1994), we expect to find only certain combinations of free and bound morphemes, if indeed there is a fixed hierarchy of functional heads. Such mixed cases thus provide evidence of a novel and special type for determining the order of functional heads. Combinations of inflected auxiliaries and aspectual particles preceding the main verb are typically found in Celtic languages. Welsh, for example, has an aspectual particle, wedi (lit. ‘after’), to express perfect aspect, but expresses Tense and (irrealis) Mood through suffixes (see Ouhalla 1988, 1991; Hendrick 1991, among others). When no aspectual particle is present, the Tense (or Mood) suffixes are attached to the main verb. See (60) = (7) of Hendrick (1991,173): (60) Cana i yfory. sing-FUT-Isg I tomorrow ‘I will sing tomorrow.’

When the aspectual particle wedi is present, Tense (or Mood) suffixes are attached to an auxiliary which precedes wedi. See (61) = (4) of Hendrick (1991, 172): (61) Byddai wedi canu erbyn saith o’r gloch. be-FUT-1isg I PERF sing by seven o’clock ‘T will have sung by seven o'clock.’

ON THE ORDER OF CLAUSAL FUNCTIONAL HEADS 67

This is expected if Tense/Mood is higher than (perfect) aspect, which in turn is higher than the main verb, as the particle in Aspyerfece prevents the main verb from reaching

Tense/Mood, thus requiring the insertion of an auxiliary to bear the Tense/Mood suffix. Welsh has a progressive aspect particle as well (yn, lit. ‘at’), which also precedes

the main verb. Again, when this is present, Tense (or Mood) suffixes appear on an auxiliary verb. See (62) = (1) of Hendrick (1991, 172): (62) Bydda i ‘n canu yfory. be-FUT-Isg I PROG sing tomorrow ‘T will sing (be singing) tomorrow.’

More generally, whenever suffixes and particles co-occur, we expect that a suffix expressing a functional head higher than the one associated with the particle will necessarily be found on an auxiliary verb (the main verb being “trapped” below the particle). The two aspectual particles of Welsh can co-occur, with the perfect aspect particle preceding the progressive aspect particle. See (63) = (6) of Hendrick (1991, 173): (63) Oedd y bachgen wedi bod yn ymlad. be-PAST the boy PERF be PROG fight ‘The boy had been fighting.’

Welsh thus gives evidence for the order of heads in (64), a conclusion already explicitly suggested in Ouhalla (1988, 71ff):* (64) T/Mood > Asppertect > ASPprogressive > V

In addition to the combination of agglutinating suffixes and (modal) particles (see n. 8), Korean illustrates another kind of mixture: the coexistence of closing and nonclosing suffixes. So, for example, the habitual suffix -kon and the progressive suffix -ko appear not to tolerate further affixation. What happens, then, is that suffixes corresponding to higher heads (Tense and speech act Mood) require the insertion of an auxiliary verb (ha- ‘do’ and) (iss- ‘be’, respectively). See (65a—b), from Song (1995, 253, 261): (65) a. Chelswu-nun cangnangam-ul mantul-kon ha-ass-ta. C.-TOP toy-ACC make-HAB do-PAST-DECL ‘C. used to make toys.’ b. Dasom-i ppalli talli-ko iss-ess-ta. D.-NOM fast run-PROG be-PAST-DECL ‘D. was running fast.’

If the derivation suggested in n. 8 (in the spirit of Kayne 1994) is correct, we indeed expect the auxiliary verb (plus the “higher” suffixes) to be to the right of the main verb with the closing suffix. This is due to the leftward movement of the XP that contains the main verb across the auxiliary verb. An entirely parallel case is found in Turkish, where neither the perfect aspect / Anterior suffix -m/ nor the progressive aspect suffix -Iyor permits further suffixation, whence the insertion of an auxiliary to bear suffixes corresponding to higher heads.

See (66):

68 ADVERBS AND FUNCTIONAL HEADS

(66) a. John hafta-ya tez-in-i bitir-mis ol-acak. (Yavas 1980, 74) J. week-DAT thesis-poss-ACC finish-PERF/ANT be-FUT ‘John will have finished his thesis (by) next week.’ b. Oku-mus ol-abil-ir. JJaklin Kornfilt, personal communication) read-PERF be-can-AOR ‘He might have read.’ c. Gid-iyor ol-abil-ir. (Jaklin Kornfilt, personal communication) go-PROG be-can-AOR ‘He can be going.’

3.5 Some remarks on prefixes, derivation, and inflection The relation between the order of (functional) prefixes on the verb and the order of the corresponding functional heads in clausal structure is rarely discussed in the literature, and remains yet to be fully understood, especially in relation to the Mirror Principle. The first thing to note concerning the order of what are taken to be verbal prefixes in various languages is the existence of (at least) two major types. In the first, exemplified by Navajo and other Athapaskan languages, the order of prefixes (at least, so called ‘conjunct’ prefixes)“ left to right appears to be just like that of suffixes, with Aspect to the left of Tense (which is to the left of subject Agreement), the only difference apparently being the position of the verb, which comes last, rather than first in the sequence: ... Asp-T-AGR-V (see Speas 1991a,b). In the second major type, exemplified by various Bantu languages, the prefixes appear to directly reflect the order of functional heads, with Agreement to the left of Tense, which itself is to the left of Aspect. (In some Bantu languages, Aspect is found suffixed to the verb root.) Both types seem to provide a challenge to the Mirror Principle, at least under a restrictive theory, such as that of Kayne’s (1994) (and Chomsky’s 1995), which only allows raising (and no lowering), and left-, but no right-, adjunction.» How, then, can these two types of prefix systems be analyzed in the restrictive theory and rendered compatible with the Mirror Principle? I can only offer some tentative suggestions. Consider, first, the case of Navajo-type languages. A possible analysis, suggested to me by Mark Baker and Lisa Travis (personal communication), is to postulate the existence of a null auxiliary, which starts the successive adjunction of lower to higher heads, with the lexical V a phonological clitic to the cluster (see also Kornfilt’s work cited in n. 43):*6 (67) ... [Aux]-Agr,-Asp-T-Agr,#V As to what are traditionally called “disjunct” prefixes in the Athapaskan literature, which precede the “conjunct” prefixes of (67) and “modify the meaning of the verb stem... in ways that can be grossly characterized as adverbial” (Rice 1991b, 99), I will tentatively assume that, much like standard cases of incorporation (see Baker 1988), they arise through the incorporation of the heads of specifiers subjacent to the head successively raising up the tree, as abstractly shown in (68):

(68) XP

ON THE ORDER OF CLAUSAL FUNCTIONAL HEADS 69

>» XP

/X HP \mRYP

/\ | PP

ZP

yY xX H Y¥ \> ZP

AS va i | / . CZ “s..L 2 ~~

NY 7

This derivation has two consequences: (1) the morphemes in the head positions of the extended projection of Z will appear in an order which is the mirror image of the underlying order (... Z-Y-X ...), aS expected under the Mirror Principle; (2) the incorporating heads of the specifiers will instead be prefixed onto Z in a non-mirror fashion, the closest to Z being the head of the lowest specifier (the one which incorporates first), and so on: H-L-Z. Some evidence for this analysis of Athapaskan disjunct prefixes may come from the following considerations, based on Rice (1991b).*’ Such prefixes (which range from six to nine, according to the language) typically comprise the following: an object

of postposition, a postposition, iterative adverbs, a distributive form, and an incorporated object in some varieties. Their nature, as well as the fact that they are “framed” by two unequivocally incorporated elements (the object of the postposition and the direct object), makes them natural candidates for an across-the-board incorporation analysis. Moreover, the limited variation found across the family, which only involves

the relative positions of the incorporated object and the adverbs (Rice 1991b, 109) or the distributive quantifier and the iterative adverb (Rice 1991b, 107), is possibly to be attributed to the different positions from which the head of the DP or QP incorporate (after possible DP-movements). No variation should instead be possible for the positions of the adverbs, if these incorporate from fixed positions. And, indeed, none is found, apparently. See Rice (1991b, 118). The order of the adverbial prefixes in Navajo is reversative (“back’ ) > semeliterative (‘again’) > iterative (which Rice does not gloss, saying that it is the most general, semantically). In the present analysis, the order should directly reflect the order of the corresponding AdvPs in specifier position. And thus it is at least suggestive that in English the most natural order appears to be back > again > seyeral times (see He had to put it back again several times).*® An “incorporation” analysis for the second major type of prefixes (that of Bantu), appears dubious, especially in relation to the Tense prefixes, for which an incorpo-

70 ADVERBS AND FUNCTIONAL HEADS

ration analysis (from the head of a subjacent specifier) looks much less natural. A more plausible analysis would consist in treating them as free morphemes (particles) clitic on the verb (as suggested, in fact, by Mchombo 1986; see Baker 1988, 468, n. 19): (69) N-ka-l4a-boomba. (ChiBemba; Giv6n 1972) Isg-FUT-PROG-work ‘Tl be working tomorrow.’

Possible support for this analysis is the fact that in some Bantu languages the “prefixes” can also appear separated from the verb, being in fact cliticized to some other constituent. See (70a—b), from Ibibio and Nomaande, respectively:” (70) a. M-ma-si-skk fi ka do. (Essien 1987, 163) AGR-PAST-HAB-still go there ‘T still used to go there.’ b. Tu-ti-ké-bul-as4 fana. (Wilkendorf 1991, 143) Ipl-NEG-FUT-HAB-1p1 read “We will not have the habit of reading (from tomorrow).’

If these conjectures are correct, “prefixes” can have three different sources. They may be heads, syntactically separate from the verb, and (phonologically) cliticized to it (Bantu languages, except those of the Shona-type); they may be suffixes on a null auxiliary (the “conjunct” prefixes of Athapaskan, and the prefixes of Shona); or they may be heads incorporated to the left of the verb from a subjacent specifier (the “disjunct” prefixes of Athapaskan). Only in the second case are they ordered in a mirror fashion with respect to the corresponding functional heads (just like the other suffixes). If so, no weakening of the Mirror Principle seems necessary, but the whole question must be investigated more closely. As a final point, I want to mention a general prediction for the order of functional heads which derives from the Mirror Principle if that 1s taken to hold of derivational and inflectional/agglutinating suffixes alike (see n. 2 above). Given that derivational suffixes are always closer to the stem than are inflectional or agglutinating suffixes, it is to be expected that where both express functional heads, the heads expressed by derivational suffixes are lower than those expressed by inflectional or agglutinating suffixes. Although this prediction has not been extensively checked in the literature, the little evidence that is available would seem to indicate that it is essentially correct. So, for example, while one finds in the same language aspectual notions encoded derivationally and mood notions encoded inflectionally (as in Central Alaskan Yup’ik; see Mithun and Ali 1996), the reverse is apparently never found. It may also happen that some languages encode certain aspectual notions derivationally and others inflectionally. But, as Bybee (1985, 101ff) notes, in such cases there is no complete freedom of expression. For example, inceptive and iterative aspects can be expressed derivationally while perfect and habitual aspects are expressed inflectionally, but not vice versa. This accords well with the fact that the latter aspects are “higher” than the former. See §3.6.°° But the whole question needs to be explored more systematically.>!

ON THE ORDER OF CLAUSAL FUNCTIONAL HEADS 71

3.6 Toward a universal hierarchy of functional heads (a first approximation) In this section, I recapitulate, for convenience, the evidence for the pairwise order of the different functional heads. As noted, such relative orders give rise, by transitivity, to a single overall order of heads (see (96)), which seems to be cross-linguistically consistent. (Many of the additional relative orders expected from the overall order are indeed attested, but I will not attempt a systematic survey. Some can be seen in the data previously discussed and in Appendix 2). Although certain interchanges in the order of suffixes are possible with argumental suffixes (reflexive/reciprocal, distributive, associative, etc.; see Fagerli 1994), and with suffixes expressing negation (see chapter 5, n. 51), 1 have come across no genuine cases in the literature in which suffixes (or particles) involving mood, modality, tense, aspect, and voice can vary their relative order. Certain suffixes may seem to have variable order with respect to other suffixes, as is the case with the Turkish -Eb/I suffix, but the suffix turns out to differ in meaning according to the

position it occupies (“possibility” in the higher one, and “permission/ability” in the lower). M O04 syeech act > M 00 yatuative

This order is motivated by the relative order of suffixes for speech act mood and evaluative mood (expressing “surprise”) in Korean. See (2), repeated here as (71):>? (71) Ku say-ka ewuk-ess-keyss-kwun-a That bird-NOM die-ANT-EVALUAT-DECL ‘That bird must have died!’ M 004 vatuative > M 00d evidential

Evidence for this order is again provided by Korean. See (3), repeated here as (72): (72) Minca-nun ttena-ss-te-kwun-yo. M.-TOP leave-PAST-EVID-EVALUAT-POLITE ‘T noticed that M. had left!’ M 00 evidential >M Od enistemic

Evidence for this order comes from Korean (see (1), repeated here as (73)), and apparently from Turkish (see (74a)), and perhaps Hixkaryana (South America) (see (74b)): (73) Ku pwun-i cap-hi-si-ess-ess-keyss-sup-ti-kka? The person-NOM catch-PASS-AGR-ANT-PAST-EPISTEM-AGR-EVID-Q ‘Did you feel that he had been caught?’ (74) a. Mary simdi ev-de ol-mali-ymis ama ben san-mi-yor-um. (Yavas 1980, 66) M. now home-loc be-must-EVID but I think-NEG-PROG-I ‘Reportedly, Mary must be at home now, but I don’t think so.’

72 ADVERBS AND FUNCTIONAL HEADS

b. Nomokyan ha-t. (Palmer 1986, 54)*3 he.come.non past.uncertain intensifier-EVID ‘He is coming (they say).’ Mod epistemic > T(P ast)

Evidence for this order is found in Korean (see again (1)); in Guyanese Creole (75a); in the Sino-Tibetan language Garo (75b); and in Oksapmin (New Guinea) (75c):*4 (75) a. Jaan shuda bin kyaan get fu gu. (Gibson 1986, 585) J. MOD gistemP AST MOD oo, MOD, oo, COMP go

‘J. should not have been able to be allowed to go.’ b. Anti-ci re’an-aha-kon. (Bybee 1985, 180) market-to go-PAST-PROBABLE ‘He probably went to the market.’ c. Go haanip di-kin-o poripti. (Lawrence 1972, 33) you person eat-PAST-PROBABLE-quotation marker say then “You probably attacked the person,” we say, then.’

T(Past) > T(Future) Converging evidence for this order comes from the order of the corresponding particles in Guyanese (and other creoles) (see (28), repeated here as (76)); in Anejom (see (77)); and from the mirror-image order of suffixes in Turkish (see (5b), repeated here as (78a)); Lezgian (see (78b)); Mongolian (see (78c)); and Malayalam (see (79)); see also (51a), from Zuni, and (4c), from Abkhaz, in Appendix 2): (76) Jaan bin gu riid. (Gibson 1986, 585) J. PAST FUT read ‘J. would have read.’

(77) Is ika aen is pu apam imran. (Lynch 1982, 119ff) PAST say he PAST FUT come tomorrow ‘He said he would come tomorrow.’ (78) a. Oku-yabil-ecek-ti-m. (Jaklin Kornfilt, personal communication) read-can-FUT-PAST-|sg ‘I was going to be able to read / I would be able to read.’ b. Im xizan.di-z jeke kiimek Ze-da-j. (Haspelmath 1993, 140ff) this-ABS family-DAT big help be-FUT-PAST “This would be a big help for the family.’ c. ox’in-d belag aw-ox bai-son. (Svantesson 1991, 200) girl-DAT present buy-FU7(irr.) be-PAST *,.. he would have bought a present for the girl.’ (79) naan innale aviDe pook-um-aayir-unnu. (Babu 1996, 12) I-NOM yesterday there go-FUT-be.PAST-PRES ‘I would have gone there yesterday.’

Additional evidence is found in the African languages Bangwa and Northern Sotho.

In Bangwa, the particles for T(Past) precede negation, whereas the particles for T(Future) follow negation (see Nguendjio 1992, 95):

ON THE ORDER OF CLAUSAL FUNCTIONAL HEADS 73

(80) a. PS; nd’ ka kwé mbé wé5. Ipl (remote)PAST NEG eat-ANT meat NEG “We had not eaten meat.’ b. P35 k3 i nkwé mbé ws. lpl NEG FUT eat-IMPERF meat NEG “We will not eat meat.’

In Northern Sotho an auxiliary marking (absolute) past can be followed by the verb in the future, to yield the (past) conditional (see (81)):°° (81) Mosadi 0-bé a-tla-réka) (nama. woman AGRs-AuxPAST AGRs-FUT-buy meat “The woman would have bought meat.’

T(Future) > Mood reais Evidence for this order is apparently provided by the Creole language Ndyuka (see (83)) and by Samoan (see (83) = (26a) of Appendix 2):°° (82) Iben o sa poi(fu) nyan ete? (Huttar and Huttar 1994, 519) 2sg PAST FUT IRR can (for) eat yet “Would you have been able to eat already?’

(83) “Ae a lava pe a seie alu atue ai lelei... but EMPH EMPH Q FUT IRR 2sg go DIR GEN eat good ‘But if you go and eat well...’ M: 004 rrealis > M O01

Evidence for this relative order comes from Ndyuka (see (82)), and from the Austroasiatic language Kammu (see (84), from Svantesson 1994, 268):°7 (84) cdo p3o pian pd mah. IRR NEG can eat food ‘{T] will not be able to eat anything.’ M 004 realis > Asp habitual

I do not know of any clear cases for such order. If the “indefinite” future particle ndwd (which precedes the habitual aspect particle n3) in Fongbe is built out of the future particle nd (“definite” future) plus an irrealis element, then a sentence like Sikd nd-wd nd da ws °Cica will eventually habitually prepare dough’ (= (101) of Lefebvre 1996) might provide evidence for Mood ,,reaii, > ASPhabituatAsp habitual > T(Anter ior ) / Asp perfect

In many cases it is not easy to tell apart anterior tense (E__R of Reichenbach 1947) from perfect aspect, so I will treat them together here in their relation to habitual aspect (but see the following section, for apparent evidence distinguishing T(Anterior) and ASP perfect and ordering the former before the latter). If “imperfect” is the default value of Asppertects Berbice Dutch Creole ((45b) in Appendix 2) is possibly interpretable as evidence for the order Asppabituai>ASP perfect:

74 ADVERBS AND FUNCTIONAL HEADS

In this creole, AsPperfect (together with various aspects) is expressed through suffixation on the verb, whereas habitual aspect is realized via a preverbal particle (das); this situation suggests that the particle fills a head higher than the head to which the verb (+ the Aspperfect Suffix) can raise. (This interpretation recalls the mesolectal Guyanese creole case mentioned in n. 25, where progressive aspect is suffixal and habitual aspect is expressed via a particle: shi doz aalweez de singing ‘she is usually always singing’). Reasons to order Asppapitua: before Asppertect are Seemingly also found in Basque,

in the Austroasiatic language Kammu, and in [sekiri, a Niger-Congo language (see (85), (86), and (87), respectively): (85) Eda-n ohi du. (Ortiz de Urbina 1989, 141) drink-PERF HAB Aux.agr ‘He usually drinks.’ (86) O kx pd mah hdoc yem kmra rdot taa kaan. (Svantesson 1994, 273) lsg HAB eat food PERF when wife come to home ‘I have usually eaten when [my] wife comes home.’

(87) O waa kd ré rén® (Omamor 1982, 110) he FUT HAB go PERF “He will have started going habitually.’

Relevant in this connection may also be Bybee’s (1985, 144) observation that in Nahuatl and Pawnee “the imperfective marker is closer to the stem than the habitual marker.” T(Anterior) > ASP perfect

Evidence for ordering T(Anterior) before Aspperfect 1s apparently offered by Kristang, a Malaccan Creole, and by the Austronesian language Banggi, even though the con-

clusion remains tentative as the relevant data and descriptions are not totally perspicuous. Kristang shows two preverbal particles, ja (related to the Portuguese adverb for ‘already’) and kaba (related to the verb kabd ‘finish’, from Portuguese acabar), which co-occur in the order ja kaba. Although he glosses them “Perfect” and “finish,” respectively, Baxter (1988, 121) says that ja is a TMA marker with a ““prior’ function” and kaba “a completive aspect marker” (p. 132), and that ja + kaba is used “to express the prior completion of the action” (p. 122). See (88a). In Bangg, the aspect marker bas, which Boutin (1991) renders as already “usually co-occurs with the perfect marker na” (p.13), in that order. See (88b):°? (88) a. Kora yo ja chega eli ja kaba bai. when Isg ANT arrive 3sg ANT PERF go ‘When I arrived he had already gone out.’ b. Sia bas na ng-korikng [m-korikng]. it ALREADY PERF ST-dry ‘It has been dry.’

Baxter (1988, 133) also notes the exact correspondence of (88a) with the following Malay sentence (89):°°

ON THE ORDER OF CLAUSAL FUNCTIONAL HEADS 75

(89) Bila gua sampai dia (sudah) habis pergi. when Isg arrive 3sg ANT PERF go “When I arrived he had already gone.’

Another case in point is possibly found in Sea Island Creole, where the perfect aspect particle done comes after the past/anterior particle been, as in she might been done cook, which Cunningham (1992, 51) renders as ‘she may have cooked [already ]’ (the done of he done ate it found in southern American varieties, and which Whitley

(1975, 94), glosses as ‘already’, may instead have been reanalyzed as an adverb). See also the case of Tokelau in (27) in Appendix 2. Asp perfect > Asp retrospecttve

While I know of no evidence for this order, evidence for the order T(Anterior) > ASP retrospective 18 found in Seychelles Creole. See (90):

(90) ZAtian fek pe maze. (Papen 1978, 362) J. PAST FUT ANT RETRO PROG eat ‘J. would just have been eating.’ Asp retrospective > Asp progressive

For this order,®! see (90), from Seychelles Creole, and the Haitian sentence (ib) of n. 30, repeated here as (91): (91) Jan fék ap kurt. J. RETRO PROG run ‘J. has just been running.’ Asp progressive > Asp prospective

Evidence for this order comes from Gungbe (see (52e), repeated here as (92), from Thai (see (20e) of Appendix 2, given here as (93)), and from (my interpretation of) Mofu-Gudur. See (7a) and the related footnote of Appendix 2, given here as (94)):° (92) Asibé nd to nd xd kéké 15, A. FUT PROG PROSP buy vélo the ‘A. will be about to buy the vélo.’

(93) Khaw kamlang ca? thuuk wicaan. he PROG PROSP PASS criticize ‘He is about to be criticized.’

(94) Ngaa kwama ya sawa na, ndaw mayal fd da 14 dola daw. of yesterday I come TOP, man thief PROG PROSP take money my “Yesterday, when I came, a thief was just about to steal my money.’ Asp prospective > Asp completrve

The “corpus” of the previous sections and of Appendix 2 fails to determine the order between completive and prospective aspect—or between completive and progressive or retrospective aspect, for that matter.

76 ADVERBS AND FUNCTIONAL HEADS

On the basis of Fula/Fulfulde (see Arnott 1970; Fagerli 1994), we know at least that completive aspect follows perfect aspect, as the latter precedes “celerative”’ aspect (marking an action performed “quickly”), which, in turn, precedes completive aspect. Thus a portion of the order of such aspectual heads remains undetermined. See (95), which is meant to represent the fact that no order can be established (on the evidence produced so far) between elements belonging to the top and bottom lines (Also recall that I have tentatively ordered durative aspect between retrospective and progressive aspect, without real evidence): (95)... ASP perfect > ASD celerative > ASP completve tae | > ASP retrospective > ASPaurative > ASPprogressive > ASP prospective |

A way to determine the missing relative orders will be discussed in the next chapter.

Following completive aspect are apparently (semel)repetitive (“again”) and iterative (“several times’’) aspects (see Fula/Fulfulde in Appendix 2). Once again, I refer to the next chapter for a more careful discussion. Putting together the pairwise relative orders considered so far, we obtain the single overall order shown in (96), at this point with certain indeterminacies and arbitrary choices (e.g., to linearize the underdetermined relative orders in (95), and regarding,

in part, the position of Voice):° (96) Mood speech act > Mood, yatative > Moodevidentiat > MOdzpistemic > T(Past) > T(Future) > Moodinreatis > ASPhabitual > T(Anterior) > Asppertect> ASP retrospective > ASPadurative > ASP progressive > ASP prospective / Mod,oot > Voice ASP celerative > ASP completive > ASP (semelyrepetitrve > ASP yterative

Though not all orders expected by transitivity from (96) are documented in our corpus, at least the many attested are compatible with it. If, indeed, the orders found in individual languages turn out not to contradict the more comprehensive order in (96) (or refinements of it), this can be no accident. The observation would, in fact, become understandable if (96) were an order available to all languages. This, in turn, is open to two interpretations. In one (the weaker), (96) would be an order determined by UG, from which individual languages “pick” their own stock of functional heads. In the other (the stronger), UG would make (96) available in its entirety to all languages. In the next chapters, I will try to argue for the stronger interpretation, suggesting that adverbs provide direct evidence for the availability of the entire hierarchy in each language.™

4

Matching and Refining the Hierarchies of Adverb Phrases and Functional Heads

If we match, left to right, the independently established hierarchies of AdvPs ((58)+(44) of chapter 1) and functional heads (96) of chapter 3), we can observe some striking correspondences (see (1) below). In many cases a transparent specifier/head relation between a certain adverb class and the right-adjacent functional head is immediately recognizable. This makes it plausible to assume that such relation should hold across the board, even in those cases where a functional head apparently finds no corresponding ad-

verb class to its left or where an adverb class finds no corresponding appropriate functional head to its right. In such cases we should not be too discouraged. It could simply be that we have failed to recognize the existence of the relevant adverb class, or of the relevant functional head (and projection). In many cases, this is what I will argue has happened. But, no doubt, other adverb classes and functional heads will have gone unnoticed. (1) [ Frankly Mood speech act | Surprisingly Mood, yamative | Allegedly Moodgyidentia | Probably Modepistemsc | once T(Past) [ then T(Future)

[perhaps Mood yreaus | cleverly ? [ usually Asppapina | already T(Anterior) [ no longer ASPpertece ? [ always ? [ ? ASPretrospective [ ? AsPaurative [ ? ASPprogressive [ ? ASPprospective

[completely ASPcompieuve | tutto ? [ well ? [ ? Voice L ? ASPceteratve |? ASD semelrepetitive [ ? ASPiteranve

Before trying to substantiate the claim that each adverb class enters into a special spec/head relation with one particular functional head, and viceversa, in §§ 4.1 to 4.3 1 will briefly sketch the analyses of mood, modality, tense and aspect that seem

//

78 ADVERBS AND FUNCTIONAL HEADS

to me most promising, and the evidence for representing each in terms of several functional heads. The literature on these topics is extensive (to use an understatement), so I will have to be selective and synthetic, yet trying to do justice to the subject matter. Surely, no easy task.

4.1 Moods and modals Mood and modality are here treated together. In part, this follows the tradition (see Palmer 1986, for a recent, particularly valuable general and typological, characterization of the two categories); in part, it reflects the fact that the same category may be expressed via mood in one language and with a modal in another, thus suggesting a close link between the two. “Mood” is traditionally restricted to modal categories (1.e. categories mostly having to do with the speaker’s opinion or attitude toward the proposition; Lyons 1977, 452) which are expressed in verbal morphology (Palmer 1986, 21). Modals instead are typically independent words (verbs, auxiliaries, or particles). In chapter 3, we have already seen some reasons to distinguish between different types of moods. In terms of both interpretation and position (in the sequence of verbal suffixes), one must distinguish indicative/subjunctive (or realis/irrealis) mood from speech act mood (declarative/interrogative, etc.). The two further categories of Evaluatives and Evidentials, of which more will be said below, are either expressed through verbal morphology, or with modal verbs or particles (see §§4.5 and 4.6). Nevertheless, they will be referred to here as “evaluative” and “evidential” moods. As to modals, we have seen evidence, in chapter 3, that the traditional interpretive distinction between epistemic and root (uses of) modals correlates with a structural distinction. Epistemic modals are located higher in clausal structure than root modals, in fact higher than T(Past)/T(Future) (and negation), apparently. Root modals, in turn, are not a monolithic class of elements. They comprise different semantic subclasses (volition, obligation, ability, permission). So one might wonder whether they all occupy the same position in the functional portion of the clause. Before considering the question, I want to briefly discuss one usage of the modals can (could), may (might) in English (and of their analogues in other languages), which is usually analysed as epistemic, and which I want to suggest should be analyzed differently (at least, different from epistemic must). In logic, epistemic modality is sometimes contrasted with alethic modality (see von Wright 1951, 1ff, and Rescher 1968, 24ff).The former is concerned with the speaker’s deductions or opinions, the latter with necessary truths (1.e., propositions that are true in all possible worlds) and with possible truths (i.e., propositions that are not necessarily false, being true in at least one possible world). See Lyons (1977, 791).! The fact that one can distinguish epistemic from alethic modality from a logical point of view is no reason to assume that they should also differ structurally. Palmer (1986, 11), for example, says that “there is no formal grammatical distinction in English, and, perhaps, in no other language either, between alethic and epistemic modality.” However, some facts from English multiple modal varieties appear to indicate that they may have to be structurally distinguished. As Brown (1992, 75)

MATCHING AND REFINING THE HEIRARCHIES 79

points out, Hawick Scots shows a “triple modal” construction, where will, marking future, is the first modal; might, marking “possibility,” the second; and could, marking “ability,” the third: (2) He’ll might could do it for you (= ‘he might be able in the future to do it for you’).

Brown also notes (p. 77) that can in he will can do it can be interpreted either in an “ability” or a “possibility” sense. If will is in T(Future), this is evidence that the “pure possibility,” or “alethic,” (uses of the) modals might and can follow T(Future),

just as root modals do. As epistemic modals precede T(Past) and T(Future), as observed in chapter 3 (n. 54) (are outside the scope of absolute tense altogether; Spellmire 1994), we have evidence that epistemic modals occupy a position distinct from, and higher than, alethic modals. In this connection, it is then interesting that a purely epistemic (i.e. non-alethic) modal like must is never found to follow will in Hawick Scots.° A comparable situation is found in Danish, where the alethic modal kunne, but

not the epistemic/evidential modal skulle, can be found following the modal vil marking the future. See (3a—b), from Vikner (1988, 10); see also Thrainsson and Vikner (1995, 51): (3) a. Der vil let kunne ga noget galt. ‘It will easily be possible that something goes wrong.’ b. *Han vil skulle have lest bogen. “He will be said to (must) have read the book.’

Vikner treats kunne as epistemic but notes that it differs from the other epistemic modals in that “it may appear in the perfect tense” (1988, n. 4), and in that it, but no other epistemic modal, can occur under an epistemic modal (p. 9ff). Treating kunne as alethic rather than epistemic (hence lower than the latter modals) may provide an account at least of its second peculiarity. These facts may thus suggest the existence of a separate modal projection of (alethic) possibility, intermediate between the epistemic and the root modal projections: Modgpistemic» « «> Mod aiethic) possibility + °° > Mod,oot.

As matter of fact, there is evidence for distinguishing a head of alethic possibility

from a head of alethic necessity. The two can co-occur in the order “necessity” > “possibility” (though not the converse). See (4), which gives evidence for the sequence. Mod pistemic +o > Mod necessity > Mod possibility > Md, cot(volition/obligation/ability/ permission).

(4) a. Perché le cose andassero a posto, questo si dovrebbe poter verificare presto.4 ‘For things to go well, it should be the case that it were possible that this happened.’ b. *Questo si potrebbe dover verificare presto. ‘It could be the case that this should happen soon.’

Consider now the root modals. These, in contrast to epistemic and alethic modalities are strictly subject-oriented. Volition, obligation, ability, and permission are properties attributed to an (animate) subject. Concerning the question raised earlier (Do the different subclasses of root modals occupy the same structural head position?), no clear answer comes from English

80 ADVERBS AND FUNCTIONAL HEADS

multiple modal varieties. In any event, they appear to enter a fixed relative scope among each other. To begin, consider volere ‘want’. It falls under the scope of alethic possibility (5a), while it takes scope over “ability” and “permission” (5b): (5) a. Gianni potrebbe voler uscire. ‘G. could want to go out.’ b. Gianni vorrebbe poter cantare. ‘G. would want to be allowed / be able to sing.’

The inverse interpretations are not possible, which suggests the relative order in (6): (6) se Modhecessity > Mod possibinty a > Mod ciition 22 > Mod, pitity/permission

To Judge from (7a—b), volere also takes scope over “obligation” (7a), although the other order, (7b)—differently from Danish, in which comparable cases are given as ungrammatical in Vikner (1988, 10): *Hun md ville ga pa indkgb ‘She must want to go shopping’—does not sound to me entirely unacceptable (in the nonepistemic reading): (7) a. Gianni vorrebbe dover prendere una decisione subito. ‘G. would want to have to take a decision immediately.’ b. ?Gianni dovrebbe volerci aiutare. ‘G. should want to help us.’

The interpretation, however, does not seem to be one of real “obligation” on the subject (‘Gianni finds himself obliged to want to help us’ ); it 1s, rather, one of general necessity (‘For everybody to be happy, it would be necessary that G. wanted to help us’).°

I assume (7b), then, to involve “(alethic) necessity” rather than “obligation,” although finer distinctions might be involved (see n. 5). This yields the relative order in (8): (8) sae Mo dhecessity > Mod, ossipmnty tm Mody otition > Mod gyigation! Mod apitity/permssion®

What about the relation among the “obligation,” “ability,” and “permission” cases of root modality? It is not entirely clear whether it makes sense, semantically, “to be obliged to be able to help,” or “to be obliged to have the permission to leave,” or “to be able to have the permission to leave.” However, the inverse scopes (“to be able to be obliged to leave”; “to have the permission to be obliged to leave”; “to have the permission to be able to leave’’’) appear even more unnatural. Some special contexts would seem to show that, if anything, the relative order among these modal notions is “obligation” > “ability/permission” (see (9)-(10)), and, perhaps, “ability” > “permission” (see the Guyanese Creole sentence (29) of chapter 3, repeated here as (11)): (9) a. ?Non vorrei dover poter risolvere I’ equazione in soli 3 minutt. ‘I would not like to have to be able to solve the equation in just three minutes.’ b. *Non vorrei poter dover risolvere |’equazione in soli 3 minuti. ‘I would not like to be able to have to solve the equation in just three minutes.’ (10) a. Ci vorremmo dover poter entrare anche noi in quel club. ‘We would like to have to have the permission to enter that club.’ b. *Ci vorremmo poter dover entrare anche noi in quel club. “We would like to have the permission to have to enter that club.’

MATCHING AND REFINING THE HEIRARCHIES 81

(11) Jaan shuda bin kyaan get fu gu. J. EPISTEM PAST NEG.ABIL PERMISS to go ‘J. should not have been able to be allowed to go.’

However, given that the evidence for ordering “ability” with respect to “permission” is rather slim, I will for simplicity consider these two notions as two different values of one and the same head (though eventually this might prove simplistic). Altogether, this yields the following strict (scope) hierarchy of modalities: (12) Modenistemic « - «> MOdneccessity > MOdpossipitity > MOdvotition > MOdoptigation > Modapuity, permission

This is perhaps not sufficient by itself to motivate five distinct modal heads, in addition to an epistemic one. However, the existence of different classes of AdvPs corresponding to the different modalities, and ordered in a similar fashion (see §4.12), can be taken as additional evidence for the hierarchy of modal heads in (12). Furthermore, we should expect that if a language were to provide morphological evidence for any such functional heads, this evidence would be compatible with the scope hierarchy in (12) (see chapter 6 for a discussion on the possible relation between semantic scope and the universal hierarchy of functional projections). This is at least partially confirmed by Turkish, for which Aksu-Kog and Slobin

(1985) give the order of suffixes: ... potential-negation-necessitative ... (see n. 4 of chapter 3), where “necessitative” “signals an obligation on the part of the subject” (Bybee 1985, 167), and “potential” seems to correspond to what was called above ‘ability.’”

4.2 Theories of tense: evidence for Vikner’s (1985) three relations theory Two main sorts of theories have been proposed in recent treatments of tense. Theories which view tenses as operators, as in Tense Logic, and those which view them as relations between temporal entities (“times”), as in the tradition stemming from Reichenbach (1947). Several conceptual and empirical arguments have been pointed

out which show the superiority of the latter over the former (see, in particular, Hornstein 1990, 92ff, and Giorgi and Pianesi, 1997 §2 of chapter 1). So, here I will not further consider the tense-as-operator approach.® In the studies stemming from Reichenbach (1947) a number of modifications have been introduced to Reichenbach’s original system.” As Comrie (1981) and Vikner (1985) independently pointed out, Reichenbach’s theory of tense has one major drawback. Taking tenses to be three-place relations of the points E, R, and $ forces him to order each point with respect to the others, thus yielding an unsatisfactory ambiguity for the future perfect (and the future of the past). A sentence such as He will have written his essay by Monday is compatible with the possibility that “the act of writing has already taken place, is taking place at the trme of speech or has yet to take place” (Vikner 1985, 88). Reichenbach’s system is forced to assign to it three distinct representations (see 9 of n. 9), and yet there seems to be no language that has a distinct morphological realization for each such representation. As Comrie (1981,

82 ADVERBS AND FUNCTIONAL HEADS

26) points out, “this suggests that the triple characterization is an artefact of the notation rather than a significant fact about language.” A more adequate representation would be one which left the relation between E and S unspecified. This can be achieved if one replaces Reichenbach’s three-place relations with two two-place relations (S_R and E_R). This solution, which is adopted by virtually all neo-Reichenbachian analyses, has the advantage of not establishing any direct relative order between E and S, thus characterizing the future perfect as vague, not ambiguous. Beyond that, there is no general consensus on what precisely the primitives of the theory are, and, as a consequence of that, on what qualifies as a possible tense. Each system makes slightly different claims in this regard. Rather than attempting a systematic comparison among the different proposals, I will mention the reasons that, to my eyes, render Vikner’s (1985) theory the closest approximation to a universal theory of tense. A few comparative remarks are found inn. 14, In Vikner (1985), three further problems with Reichenbach’s original system are pointed out:

1. The same basic tense (future) is given two distinct representations (see 4 and 5 of n. 9). But languages do not seem to make a distinction in the future comparable to that between the present perfect and the simple past. 2. The system includes a form (the “posterior future’; see 7 of n. 9), for which there is no clear linguistic evidence.!° 3. It makes no room for something which is a tense form: the future perfect of the past (.. . “he would have worked”). Vikner suggests that such defects can be remedied by introducing another reference point (which 1s needed for the future perfect of the past; see his discussion of She promised in November that they would have received her paper by the first day of term, and the similar proposals of Comrie 1981 and Lindstedt 1985, 33ff) and by constraining the system to three strictly binary and oriented relations (between two times). The two times may either coincide (the unmarked option; Vikner 1985, 94) or be in a unique precedence or subsequence relation to each other: (1) E coincides with, or precedes, R, (let us call it for convenience T(Anterior)); (2) R, coincides with, or follows R, (for convenience T(Future)); (3) R, coincides with, or precedes, S (for convenience T(Past)).!! The actual tenses arise from the combinations of the different values of the three basic two-place relations, as shown in (13):

(13) a. R,,S; R,,R.; E,R> = present (works) b. Ry.5; R,;_R.; E.R, = future (will work) c. R,_S; R,,R.; E,R,= past (worked) d. R,,S; R,,R,.; E_R,= anterior (has worked) e. R,_S; R,,R,; E_R, = anterior of the past (had worked) f. R,_S; R,_R.; E,R, = future of the past (would work)

g. R,,8; R,_R,; E_R, = anterior of the future (will have worked) h. R,_S; R,_R,; E_R, = anterior of the future of the past!” (would have worked)

The system is very restrictive. It claims that the eight tenses of (13a—h), which exhaust all the possible combinations of the three basic relations, are the only possible

MATCHING AND REFINING THE HEIRARCHIES 83

tenses to be found across languages.!? In chapter 3, we saw that certain languages appear to bear this system on their sleeves, so to speak. Seychelles Creole, for example, showed a distinct particle for each of the three relations (in fact combining T(Anterior), T(Future) and T(Past)). See (48) of Appendix 2. This direct mapping from abstract tenses to morphemes (which extends to English -ed, will, and -ed) must certainly count as another virtue of the system (see Hornstein 1990, §3.7). In what follows, [ will adopt Vikner’s (1985) theory essentially as it is, only combining it with one important suggestion of Giorgi and Pianesi’s (1991, 1997) (see also Zagona 1988, 70ff)—-namely that each (here, strictly binary, and “oriented”) relation corresponds to a functional head of the extended projection of V; a separate T°: T(Anterior), T(Future), and T(Past), with the values “default” (when the two times coincide) and “marked” (when they do not). The three tenses are 1n a particular scope relation to each other (with T(Anterior) embedded under T(Future), itself embedded under T(Past), a fact that remains to be derived):'4

(14) [...[T1 (Past)... [T2 (Future)... [ T3 (Anterior)... V]] |] 4.3. “Lexical” and “grammatical” aspect Two quite different things fall under the term “aspect,” which are often, though not always, kept separate in the literature. One is the internal structure of the event, or situation, as lexically expressed by the predicate and its arguments: whether it has a beginning or end, internal stages, etc. Vendler’s (1967) classical typology distinguishes among “activities” (run, push a cart), “states” (know, desire), “accomplishments” (run a mile, build a house), and “achievements” (reach the top, find a wallet).!° The other refers to the particular way in which the speaker presents the event, or situation, through grammatical means—for example, as terminated (through the perfect aspect: John has run a mile); as on-going (through the progressive aspect: John was running a mile); as habitual (through the habitual aspect: John used to run a mile); and so on.!6 The terminology employed to distinguish these two “senses” of “aspect” var-

ies. Smith (1991), for example, calls the first “situation aspect” and the second “viewpoint aspect.” Following Dahl (1985) and others, I use here “lexical” and “grammatical” aspect, respectively, even if the first term may not be entirely appropriate.!’ Of course, to distinguish these two kinds of “aspect” is not to deny the existence of relations between them. So, for example, “progressive” aspect, which marks an on-going activity, 1s incompatible with states (see *He is knowing that we like him). Similarly, so-called completive aspect (which marks the completion of all internal stages up to the attainment of the natural end point) is only applicable to “accomplishments.” Interesting as they may be, I will generally leave such questions to the side here, focusing rather on the several types of “grammatical” aspects found in the languages of the world. They will be discussed separately next, together with the particular adverb class that seems to correspond. In the next section, I begin to discuss the matching of adverb classes and functional heads, from left to right (top to bottom).

84 ADVERBS AND FUNCTIONAL HEADS

4.4 Speech act adverbs and speech act mood By speech act mood I have indicated those grammatical means, most often encoded as affixes on the verb, that mark the basic illocutionary force of a sentence. Many languages distinguish a declarative from an interrogative and imperative mood. Others make finer distinctions (see Hengeveld, forthcoming b, §2.1).!8 Speech act mood, when expressed via a suffix, is generally the outermost (only followed by suffixes marking subordination, if any; see the case of Korean briefly discussed in chapter 3 (§3.1, n. 3). I took this to indicate that speech act mood is the highest head of the IP “space” (Grimshaw’s 1991 extended projection of V).'? One might also wonder whether speech act mood shouldn’t rather be part of the complementizer “space” (if a sharp distinction can be made between the two “spaces”’). Rizzi (1997) presents evidence for positing a head, within the complementizer “space,” which marks the illocutionary force of the sentence, and which is distinct from, and higher than, other C heads. However, it is dubious that Mood”, ecch act and Rizzi’s Force® should be identified. If such adverbs as frankly, honestly, sincerely, and so on, which qualify the speaker’s act of declaration, are plausibly to be taken as specifiers of the speech act mood head (they precede evaluative adverbs—see chapter 1—just as the speech act mood head precedes the evaluative mood head), then we have some indication that the two heads are distinct. The reason is that the illocutionary force C° precedes topicalized and focalized phrases (Rizzi 1997), yet such speech act adverbs seem to be able to follow them. See (11): (15) Di questo, A NESSUNO francamente potrei parlare. ‘About this, to nobody (focus) frankly I could talk.’

The fact that they can also precede topicalized and focalized phrases, if separated by a pause (Francamente, di questo, ANESSUNO potrei parlare ‘Frankly, about this, to nobody (focus) I could talk’) may indicate that they can also move to the spec of Rizzi’s ForceP (just as evaluative and epistemic AdvPs appear to be able to move to the complementizer “space” in French; see chapter 1, n. 53 and relative text).”°

4.5 Evaluative adverbs and evaluative mood In the logic tradition, a class of “evaluative” modalities is recognized, which Rescher (1968, 24ff) glosses as “It is a good/perfectly wonderful/bad thing that p” (see Palmer 1986, 12ff). Across languages, such modalities are either expressed by bound morphemes (suffixes) or by free morphemes (modals or particles). They do not affect the truth of the proposition, but rather express the speaker’s (positive, negative, or other) evaluation of the state of affairs described in it. We have already seen the case of the Korean suffix -kwun-, expressing “surprise” (see (2) and (3) of chapter 3, repeated here as (16a—b). Palmer (1986, 120) cites the Australian language Ngityambaa’s “good job” / “bad job” particles Grom Donaldson 1980, §1.5.4), which “express approval and disapproval of certain ob-

MATCHING AND REFINING THE HEIRARCHIES 85

served facts” by the speaker (see (17a—b), and the Menomini verbal suffixes -asah,

-apah, which “express failure of expectation, surprise or disappointment” (see (18a—b)):! (16) a. Ku say-ka cwuk-ess-keyss-kwun-a. that bird-NOM die-PAST-EPISTEM-EVALUAT-DECL ‘That bird must have died!’ b. Minca-nun ttena-ss-te-kwun-yo. M.-TOP leave-PAST-EVID-EVALUAT-POLITE ‘T noticed that M. had left!’

(17) a. Mandan-gul-dhii-ndu wana:y niyiyi. good job-lOBL-2NOMNEG say-PAST ‘Good job you didn’t tell me.’ b. Ga:mbadayana-nhi. bad job go-PAST ‘Bad job (she) came!’

(18) a. Piasah. ‘So he is coming after all! (despite our expectation to the contrary).’ b. Piapah. “But he was going to come! (and now it turns out that he is not).’

In connection with Palmer’s (1986, 120) remark that “indeed there are closed systems of evaluative adverbs in many languages that are not to be discussed here,” I suggest (what is perhaps now obvious) that such (English) adverbs as (un)fortunately, luckily, regrettably, surprisingly, strangely/oddly (enough), (un)expectedly, and so on (and their equivalents in other languages) are generated in the specifier position of an evaluative mood head taking the different values observed. It is significant that, just as the evaluative suffix of Korean is ordered between the speech act mood and evidential suffixes, evaluative adverbs were seen in chapter 1] (§ 1.2) to be ordered between speech act and evidential adverbs.

4.6 Evidential adverbs and evidential mood As anticipated in chapter 3 (n. 13), in many languages a verbal affix, or a (modal) auxiliary, or a particle is used to express the type of evidence the speaker has for his/ her assertion.?* Some languages have quite elaborate evidential systems, making as many as five or six distinctions, according to whether the speaker has visually witnessed the situation described, has had only auditory evidence, or sensory evidence of some other kind; has heard someone else’s report, had “revelative” evidence (a dream), has evidence from his own previous experience, etc. For a discussion of different systems, I refer to the works quoted in n. 13 of chapter 3, and especially to Chafe and Nichols (1986), Palmer (1986, 66-76), and Willet (1988). Other languages simply make a distinction between direct evidence (usually unmarked) and “reported” or “hearsay” evidence (“quotative”’ evidentiality). This is the case of the Danish modal skulle (19) and the German modal sollen (20): (19) De skall ville bygge et hus. (Vikner 1988, 9) “They are said to want to build a house.’

86 ADVERBS AND FUNCTIONAL HEADS

(20) Bei den Unruhen soll es bisher vier Tote gegeben haben. (Palmer 1986, 53) ‘So far four people are reported killed in the disturbances.’

In Romance, the “conditional mood” can have a similar “quotative” usage. See (21), from Italian:”° (21) Hai sentito il telegiornale? Ci sarebbe stato un rapimento importante. ‘Have you heard the news? There would have been (= it is said there was) an important kidnapping.’

Other languages instead have evidential particles. See the Basque case reported in (58b) of chapter 3, and the Hidatsa cases given in (39a—b) of Appendix 2. In those languages lacking evidential affixes, modals, or particles, the same basic distinctions can be expressed by periphrasis (it is said that... , etc.), or by evidential adverbs, which I take to be generated in the specifier position of the Evidential Mood Phrase. Plausible candidates for this class of adverbs (in English) are allegedly, reportedly, apparently, obviously, clearly, evidently, and so on.*4 Again, just as the Korean evidential suffix is ordered between the evaluative and the epistemic suffixes (see (1) and (3) of chapter 3), so evidential adverbs were seen in chapter 1 to be ordered between evaluative and epistemic adverbs. So precise a correspondence can hardly be accidental. In fact, it falls out from the restrictive X-bar theory of Kayne (1994) if we set adverbs in a spec/head relation with the semantically corresponding functional heads.

4.7 Epistemic adverbs and epistemic modals While, as noted in §4.1, alethic modality expresses necessary or possible truths, and root modality “obligation, permission, volition or ability on behalf of an agent which usually, but not necessarily, is expressed by the . . . subject of the sentence” (Platzack 1979, 44), epistemic modality expresses the speaker’s degree of confidence about the truth of the proposition (based on the kind of information he/she has). In English, for example, the epistemic uses of must express a confidence stronger than that expressed by the epistemic uses of should (but still less than absolute): (22) a. John must be home, now. b. John should be home, now. As many people have observed, the same lack of confidence on the part of the speaker

can also be expressed by such “speaker-oriented” or “epistemic” adverbs as probably, likely, presumably, supposedly, and so on (John is probably home, now).*© This

is natural if such adverbs are generated 1n the specifier position of the projection headed by epistemic modals. In fact, epistemic modals and epistemic adverbs pattern alike with respect to certain properties. For example, Jackendoff (1972, 84, 102ff)

notes that neither “feel[s] comfortable in questions” (see the ungrammaticality of *Must it be five o’clock? and *Did Frank probably beat all his opponents ?), concluding that “if epistemic modals are treated like speaker-oriented adverbs by the semantic component, this restriction will follow automatically” (p. 103). (I take such unexceptional cases as Must anyone who registered be a sophomore’, noted in Brennan (1993, 69) as instantiating alethic necessity rather than epistemic modality.)

MATCHING AND REFINING THE HEIRARCHIES 87

Secondly, as seen in §3.2, epistemic modals are higher than root modals and take the latter necessarily in their scope. If epistemic adverbs are generated as specifiers

of the same projection, the fact that they necessarily take scope over root modals even when the latter precede them, as in (23), can then be straightforwardly derived (see T. B. Ernst 1991, 754 ff; 1992, §4 for similar conclusions). What apparently counts

for (this type of) scope relations is the representation before (head-)movement (or after “reconstruction”’): (23) John must probably give his money back by tomorrow.

4.8 Time adverbs and T(Past), T(Future) The matching between temporal adverbs and T(Past), T(Future) would appear to be straightforward, especially in view of the recurrent observation that languages lacking overt tense distinctions often resort to temporal adverbs to locate the event in time.?’ Analogously, in those languages (such as the Indoeuropean languages) that make no “metrical” (or “remoteness’’) distinctions in their tense system (i.e., do not distinguish a past located earlier in the same day from one located within the previous day, or weeks/months before, or in the remote times) the same distinctions can be expressed by adding explicit temporal adverbs.78 No such simple matching seems possible, however. Adverbials like ieri “yesterday’, domani ‘tomorrow’, due mesi fa ‘two months ago’, /’anno scorso ‘last year’, and so on cannot appear between epistemic adverbs and lower adverbs (in fact, cannot appear within the IP functional “space” at all): (24) a. *Gianni sara (probabilmente) domani (stupidamente) licenziato. ‘G. will (probably) tomorrow be (stupidly) fired.’ b. *Gianni non ha (probabilmente) ieri (mai) dormito. ‘G. didn’t (probably) yesterday (ever) sleep.’

Such adverbs can only appear sentence initially or sentence finally positions, which in chapter 1 were seen to be reserved to “adverbs of setting” and “circumstantial adverbials,” respectively.”” The only temporal adverbs admitted in this position are the purely deictic ones: future and past, allora ‘then’, ora ‘now’, and, perhaps, un tempo, una volta ‘once’. As (25) to (28) show, their canonical position seems to be between epistemic adverbs and forse ‘perhaps’. The judgments are rather delicate as they must abstract from the parenthetical, or “comma intonation,” uses, of the adverbs; from the reading of allora = ‘consequently’; and from the “focusing” usages of probabilmente and jorse, which form a constituent with the following word: (25) a. Leic’era probabilmente anche allora stata. ‘She had probably then been there too.’ b. *?Lei c’era anche allora probabilmente stata. ‘She had then probably been there too.’

(26) a. Era allora forse stata fortunata. ‘She had then perhaps been lucky.’ b. *?Era forse allora stata fortunata. (ok with forse allora a constituent) ‘She had perhaps then been lucky.’

88 ADVERBS AND FUNCTIONAL HEADS

(27) a. Ci sara probabilmente anche allora stato. “He will probably then have been there too.’ b. *?Ci sara anche allora probabilmente stato. “He will then probably have been there too.’

(28) a. Sara allora forse finalmente rifatta. ‘It will then perhaps be finally restored.’ b. ??Sara forse allora finalmente rifatta. ‘It will perhaps then be finally restored.’

In chapter 1, ora ‘now’, was seen to canonically occur in between epistemic adverbs

and forse ‘perhaps’. While its meaning makes it the obvious candidate to enter a specifier/head relation with “present tense,” this tense has no specific structural position in Vikner’s system reviewed in §4.2. It results “compositionally,” when the time points related by T(Anterior) (E and R,), T(Future) (R, and R,), and T(Past) (R, and S) coincide (1.e., have the “default” values). Nonetheless, it is plausible to view the “present” to correspond to R,,S, the default value of T(Past), provided that the lower T°s also have the default value. If so, the distribution of ova seen in chapter 1 is no longer surprising.*® What remains to be seen is whether binary tense systems (where the Present has the same morphological realization as Past versus Future, or as Future versus Past; see Comrie 1985, 48ff; Frawley 1992, 360ff) can be accommodated naturally within the framework adopted here.

4.9 “Perhaps” and irrealis mood In chapter 1, we saw that forse ‘perhaps’ can co-occur (somewhat redundantly) with probabilmente ‘probably’, in the order probabilmente forse rather than vice versa. Similar judgments seem to hold for Dutch waarschijnlijk ‘probably’ and misschien ‘perhaps’ (Jeannette Schaeffer, Marcel den Dikken personal communications). Consistent with this order is the observation made in the previous section (and

in chapter 1) that forse follows allora/ora, which, in turn, follow probabilmente. Forse also differs from probabilmente (and the other epistemic adverbs) in that it can occur in questions, as Bellert (1977, 344) observed with respect to English perhaps: (29) a. Gianni é forse gia stato qui? ‘Has G. perhaps been here before?’ b. *Gianni é probabilmente gia stato qui? ‘Has G. probably been here before?’

Forse, perhaps, misschien, and so on cannot then be assigned to the same class of (epistemic) adverbs to which probabilmente belongs. I tentatively suggest that the adverb corresponds to Irrealis Mood. The fact that it comes after time and epistemic adverbs thus meshes well with the fact that Irrealis Mood follows Epistemic Modality, and T(Past) and T(Future), as observed in chapter 3. Hidatsa has a mood, called by Matthews (1965) “perhaps” mood, which is “used when the speaker doesn’t know if the proposition is true and doesn’t think the addressee knows either’ (Sadock and Zwicky 1985, 168). See also the “dubitative” suffix of the Australian language Thargari, which Klokeid (1969, 42f) renders as ‘perhaps’.

MATCHING AND REFINING THE HEIRARCHIES 89

4.10 “(Not) necessarily/possibly” and alethic modals In discussing alethic modality earlier, I suggested that it is distinct from both epistemic and root modality, in fact intervening between the two.*! Example (2) indicates that it is lower than T(Future), but says nothing about its relation to Mood,,,cayIf forse is in Spec, Mood., cai; and necessariamente /non necessariamente in Spec, Modnecessitys SPEC, MOd possibunty, Fespectively, the following sentences provide evidence for the order Mood ireatis > Modhecessity / Mod possipitity:

(30) a. Saranno forse necessariamente riammessi. “They will perhaps necessarily be readmitted.’ b. *Saranno necessariamente forse riammessi. ‘They will necessarily perhaps be readmitted.’

(31) a. Iricchi non sono forse necessariamente felici. ‘The rich aren’t perhaps necessarily happy.’ b. *[ ricchi non sono necessariamente forse felici. ‘The rich aren’t necessarily perhaps happy.’

As T(Past) and T(Future) precede Mood,,,.a1:,, we expect, by extension, that (future and past) allora ‘then’ would also precede (non) necessariamente, a correct prediction, apparently: (32) a. Neanche loro saranno allora necessariamente dalla vostra parte. ‘Not even they will then necessarily be on your side.’ b. *Neanche loro saranno necessariamente allora dalla vostra parte. ‘Not even they will necessarily then be on your side.’ c. [russi non erano allora necessariamente comunisti convinti. “The Russian weren’t then necessarily convinced communists.’ d. *[ russi non erano necessariamente allora comunisti convintt. ‘The Russian weren’t necessarily then convinced communists.’

4.11 Subject-oriented adverbs and root modals The special relation between “subject-oriented” adverbs and (English) root modals is explicitly recognized in the literature. Zubizarreta (1982, 35ff, 123ff; 1986) analyzes both as adjunct predicates which assign an additional (adjunct) 6-role to the subject: “As in the case of S-adverbs [’subject-oriented” adverbs], we then expect that the “orientation” of modals may change under passive” (1982, 129). In a way parallel to (33), where the adverb retains its orientation on the subject when this changes, root modals also appear to retain their orientation on the subject, as seen in (34), where the permission, or obligation is once on the doctor, and once on John (see Roberts 1985a, 50; Travis 1988, 304):*? (33) a. Joe intentionally has seduced Mary. (Jackendoff 1972, 82) b. Mary intentionally has been seduced by Joe. (34) a. The doctor may/must examine John. b. John may/must be examined by the doctor.

90 ADVERBS AND FUNCTIONAL HEADS

Another property that subject-oriented adverbs and root modals share is the fact, observed in Roberts (1985b, 183) (see also Roberts 1985a, 5O0ff; 1987, 186ff), that neither can occur in middles: (35) *The book sold voluntarily. (36) a. *Arabic can read easily. (with root interpretation) b. *Dinner must serve best at 8. (with root interpretation) Root modals were seen above (§4.2) to enter a particular scope hierarchy, with modals of ability in the scope of modals of obligation, themselves in the scope of modals of volition. Interestingly, different classes of AdvPs exist which seem to correspond to the different classes of root modals; and which seem to display a corresponding relative order. See (37)-(38):*3 (37) a, Gianni si presenta volentieri obbligatoriamente al posto di polizia. ‘G. willingly obligatorily goes to the police headquarters.’ b. *Gianni si presenta obbligatoriamente volentieri al posto di polizia. ‘G. obligatorily willingly goes to the police headquarters.’ (38) a. Gianni inevitabilmente lascera goffamente cadere la tazza. ‘G. will inevitably clumsily drop the cup.’ b. *Gianni goffamente lascera inevitabilmente cadere la tazza. ‘G. will clumsily inevitably drop the cup.’

Examples (37a—b) indicate that “volition” adverbs precede “obligation” adverbs, and (38a—b) that “obligation” adverbs precede “ability” adverbs. All in all, there is some evidence that the different classes of root modal adverbs mentioned in n. 33 enter into a systematic relation with the three distinct root modals isolated above, thus justifying the postulation of three distinct root modal projections, in the order: .. . > Mody otition > Mod opiigation > Mod pitity/permission a There is some indication that root modals may be lower than some aspectual heads.

We saw previously (chapter 1, $1.4) that volitional adverbs like intentionally appear to follow frequentative aspect adverbs like twice (often, etc.). Likewise, they appear to follow habitual and repetitive aspect adverbs (Gianni lasciava di solito/di nuovo intenzionalmente cadere tutte le loro offerte ‘G. usually/again intentionally let all their offers drop’ versus *?Gianni lasciava intenzionalmente di solito / di nuovo cadere

tutte le loro offerte). Analogous facts are found with deontic modal adverbs, like obbligatoriamente. Furthermore, ability/permission root modals appear to occur lower than progressive aspect in Turkish (see Yavas 1980, 66). Here, I will not try to determine their exact Jocation in the hierarchy, which requires a separate treatment.

4.12 Habitual adverbs and habitual aspect As seen here, aspectual heads are invariably lower than mood, (absolute) tense, and (non-root) modal heads, and enter into a fixed sequence. From this section on, I will consider each aspect in turn (and the adverbial class that seems to correspond to it), beginning with habitual aspect, which is the “highest,” apparently (see chapter 3, end of §3.5, and n. 5Q).

MATCHING AND REFINING THE HEIRARCHIES 91

Comrie (1976a, 27ff) defines habitual aspect as describing “a situation which 1s

characteristic of an extended period of time” (in fact, characteristic of the whole period), and explicitly distinguishes it from iterative, or frequentative, aspect, which indicates “the mere repetition of a situation.” See also Dahl (1985, 97).*> The adverb class corresponding to this aspect plausibly comprises such adverbs as usually, habitually, customarily, generally, regularly, etc. Dahl (1985, 97), in fact, explicitly says that “the cases where HAB[itual aspect] is typically used are those in which the adverb usually is possible in English.” In addition to the near-paraphrase relation between habitual morphology and habitual adverbs (see Last year, I used to wake up at 7; Last year, I usually woke up at 7), the two display a parallel position in the head and adverb hierarchies. Habitual aspect appears to precede frequentative aspect (as shown, for example, by the order of suffixes in the Papuan language Yareba, in (39a), and of particles in Rapanui (Austronesian), in (39b)). Similarly, habitual adverbs precede frequentative adverbs (see (40)):°° (39) a. yau-r-edib-eb-a-su. (Weimer 1972, 61). sit-CM-FREQ-HAB-PRES-3sgMasc “He (habitually and repeatedly) sits down.’ b. Pura vara tu’u mai a Nau. (Du Feu 1996, 162) HAB FREQ come toward Pers.sing. N. ‘N. usually comes here.’

(40) a. Mario é di solito spesso costretto a rimanere a casa. *M. is usually often obliged to stay home.’ b. *?Mario é spesso di solito costretto a rimanere a casa. *M. is often usually obliged to stay home.’

In certain languages (e.g., Italian), a habitual reading is available even in the absence of special (habitual) morphology on the head, or of a habitual adverb, provided that the verb is in a non-perfect form. See (41) and Bertinetto (1994a,b) for discussion.

(41) a. L’anno scorso, mi alzavo alle 7. ‘Last year, I (usually) woke up at 7.’ b. Quest’anno, mi alzo alle 7. ‘This year, | (usually) wake up at 7.’

In such cases, the verb (in the past “imperfect” and in the present) perhaps can be assumed to raise and check the marked habitual feature in the Aspyapitya1 head.?’

4.13 Repetitive/frequentative adverbs and repetitive/frequentative aspects (I) In the literature, such terms as “frequentative,” “iterative,” and “repetitive” are used sometimes as synonyms (though maybe applied by different authors to different things) and sometimes 1n opposition to one another to refer to different kinds of things. So, care must be taken to understand what is referred to each time. Beyond the terminological problem, there is a question of substance. Languages seem to make distinctions in what we can refer to as “repetition.” Even though in certain languages the same form may signal that an action is repeated on a single occasion or on different occasions (Bybee and Fleishman 1985, 160; Comrie 1976a, 43,n. 2), other languages distinguish these two types using different morphological

92 ADVERBS AND FUNCTIONAL HEADS

affixes. So, for example, the Austronesian language Sobei distinguishes between performing a certain action repeatedly, several times, or often and performing it again.*® In fact, the corresponding adverbs can co-occur, with the order again > often. Both types of repetition may quantify over events in which a certain process is involved, or over the process itself, quantification over events being located higher than quantification over the process.This 1s particularly clear with the corresponding AdvPs (which, I take, as expected by now, to be in the specifier position of the respective aspect phrases).

Example (42) shows that the same “frequentative” adverb, spesso ‘often’, can occur in two different positions (demarcated by the adverb gid ‘already’). Example (43), more revealingly, shows that both positions can be simultaneously filled:°? (42) a. (Quando troviamo qualcosa) questa é spesso gia stata scoperta da qualcuno. ‘(When we find something) this has often already been discovered by someone.’ b. Questa proprieta é gid stata scoperta spesso, negli ultimi cinquant’ anni. ‘This property has already been discovered often, in the last fifty years.’ (43) a. Gianni, saggiamente, spesso esce con la stessa persona spesso. ‘G., wisely, often dates the same person often.’ b. Gianni raramente esce con la stessa persona spesso.*® ‘G. rarely dates the same person often.’

This is not to say that there are two spesso (often, souvent, etc.). Rather, it seems plausible to take the same lexical item to be generable in two distinct (scope) positions.*! That the two positions of spesso, etc. are specific scope positions (unlike what

free adjunction to different phrases would make one expect) 1s shown by the fact that each is fixed relative to other adverb positions. So, for example, the “higher” spesso follows habitual adverbs (see (40)), and precedes gia (see (42a) versus * é gia spesso Stata scoperta da qualcuno), while the “lower” spesso follows the much lower adverb bene, and at least some of the complements (Gianni parlava bene (*spesso) di noi (spesso) con sua madre ‘G. would talk well about us often with his mother’ ).*7

Although I do not know if any languages use the same affix twice, to mark a “higher” and a “lower” (frequentative) aspect, I will nonetheless assume the existence of two separate aspect phrases corresponding to the two scope positions occupied by spesso and other such terms, calling them, for ease of reference, the higher ASP frequentanvey 20d the lower ASsPfrequentanvea- A potential consequence of this assump-

tion is that some language might give the illusion of having a “frequentative” aspect ordered differently (with respect to another aspect) than some other language. West Greenlandic may be a case in point (see the discussion in Fortescue 1984, 286).* An analogous double position (and scope) appears to be available for adverbs expressing repetition on a single occasion (di nuovo, nuovamente, ancora, and others ‘again’), and, I take, for the corresponding aspect phrases. See (44): (44) Gianni ha di nuovo battuto alla porta di nuovo/ancora. ‘G. again knocked on the door again.’

The leftmost di nuovo quantifies over the event (of knocking on the door, perhaps many times), while the rightmost quantifies over the act itself of knocking. (See also Stechow’s 1996 two-position analysis of repetition wieder ‘again’ in German.)

MATCHING AND REFINING THE HEIRARCHIES 93

These adverbs, which I will take to be in the specifier position of an ASPyepetitrve) and ASP,epetitivea» Seem to be at the immediate left of the adverbs in the specifier position of ASPfrequentative) ANd ASPérequentativean, Tespectively. See (45) and the ungram-

maticality of (46a—b) with a different order:“ (45) ?Gianni di nuovo raramente vede la stessa persona ancora spesso. ‘G. again rarely sees the same person again often.’

(46) a. *Gianni raramente di nuovo vede la stessa persona. ‘G. rarely again sees the same person.’ b. *Gianni raramente parla spesso di nuovo con la stessa persona. ‘G. rarely speaks often again with the same person.’

4.14 “Quickly/rapidly” and celerative aspect (I) As noted in Travis (1988, 292), an adverb like guickly (also rapidly, etc.) may quantify over the event, as 1n (47a) (“John was quick in..”’), or over the process, as in (47b) (“John did it in a quick way”) (in (47a), but not in (47b), he may actually have lifted the arm with a slow movement):*° (47) a. John quickly lifted his arm. b. John lifted his arm quickly. The different interpretation appears once again to be a function of the (scope) position occupied by the adverb, the leftmost being the one associated with quantification over events. This position seems to be contiguous to that of Frequentative(I) (and Repetitive(I)) adverbs; in fact, to their nght and to the left of gia ‘already’. See (48)—(50): (48) a. Gianni ha di nuovo rapidamente cambiato opinione. ‘G. has again quickly changed his mind.’ b. ??Gianni ha rapidamente di nuovo cambiato opinione. ‘G. has quickly again changed his mind.’

(49) a. Gianni ha spesso rapidamente cambiato opinione. "G. has often quickly changed his mind.’ b. *?Gianni ha rapidamente spesso cambiato opinione. ‘G. has quickly often changed his mind.’

(50) a. Se Mario mi avra rapidamente gia avvisato per le due... ‘If M. will have quickly already warned me by two o’clock.. .’ b. *Se Mario mi avra gia rapidamente avvisato per le due... ‘If M. will have already quickly warned me by two o’clock .. .’

A lower position hosting “quickly” adverbs will be discussed in §4.25, in relation to what is known in the africanist tradition as “celerative” aspect (Arnott, 1970, 356; Fagerli 1994, 36ff), a particular verbal morphology signaling that the action has been performed quickly. As with the repetitive and frequentative aspects discussed previously, I will assume that corresponding to the higher guickly there is a higher aspectual projection (ASP ccterative wy) .

94 ADVERBS AND FUNCTIONAL HEADS

4.15 “Already” and T(anterior) Abstracting away, for the moment, from the negation position occupied by mica (to which I return in chapter 5), the next adverb class down 1s that represented by gia ‘already’ (see chapter 1, §1.1). As noted in the literature, its core meaning is one of temporal priority (see, in particular, Hornstein 1977, 547ff; Michaelis 1991); in fact, one of pre-

cedence with respect to a reference time. See the following discussion of (51)-(52). This makes it plausible to locate it in the specifier position of the lowest TP (TP. nterior)> Which marks the relation E_R, (i.e., Event time precedes Reference time). Consider (51a—b):

(51) a. Haven’t we already met? b. Last Christmas, hadn’t they already met?

The contribution of already appears minimal. Example (Sla—b) mean the same as “Haven’t we met?” and “Last Christmas, hadn’t they met?,” with the added presupposition that the encounter is located before some reference time (= now and last Christmas, respectively). As Hornstein (1977) also observes, in a complex sentence of the form Sentencel when Sentence?2, “The adverb already forces a priority reading for the event expressed in the sentence in which it is found” (p. 547). See (52a—b): (52) a. John had gone surfing when Harry had gone swimming. b. John had already gone surfing when Harry had gone swimming.

While in (52a) the events expressed by the matrix and the when clause can be contemporaneous, 1n (52b) the presence of already 1n the matrix clause forces the event expressed by it to precede the event expressed by the when clause. Whence, also, the ungrammaticality of *John had already gone surfing while Harry had gone swimming, where when is replaced by while, which is only compatible with the two events being contemporaneous. Fortescue (1984, 278) glosses the affix -riir- of West Greenlandic “which indicates an action completed prior to some reference point” with “already” (also see chapter 3, n. 24, and relative text).* (53) Niri-riir-pugut. eat-already-1plIND “We have/had already eaten.’

4.16 “No longer” and terminative aspect The next head down, in the hierarchy seen in chapter 3, is ASPpertecr- The next adverb

down in the hierarchy of AdvPs seen in chapter 1 1s (non...) piw ‘not any longer / no longer’. It is unclear, however, whether the latter is in any systematic relation with the former. Certain classes of verbs, when used in the perfect, do seem to imply that the situation described no longer obtains (see (54a—b)), but this is not true with others (see (54c—-d)):47

MATCHING AND REFINING THE HEIRARCHIES 95

(54) a. Gianni ha amato molto Maria (— non l’ama pit). ‘G. has loved M. a lot’ (- he no longer loves her). b. Gianni ha mangiato (—-> non sta pill) mangiando). ‘G. has eaten’ (—> he is no longer eating).

c. Gianni é partito (4 non parte pit). ‘G. has left’ (# he is no longer leaving). d. Gianni ha raggiunto la vetta (6 non la raggiunge pit). ‘G. ha reached the summit’ (4 he no longer reaches it).

(Non...) pit, more than with perfect aspect, appears to enter a periphrastic relation with what is sometimes referred to as “terminative” (or “cessative”) aspect (Binnick 1991, 204; Frawley 1992, 321), which characterizes a situation as having reached an end point, though not necessarily the natural end point. This aspect is encoded affixally

in certain languages (West Greenlandic; Fortescue 1984, 283; Hixkaryana— Derbyshire 1985, 225), with particles in others (Ewe; Ameka 1988, 2041f), and via an aspectual verb in others still (English or Italian; Freed 1979):* (55) a. Gianni ha smesso di amare Maria (= Gianni non ama pit: Maria). ‘G. has stopped loving M’ (= G. no longer loves M.). b. Gianni ha smesso (cessato) di cantare (= Gianni non canta piu). ‘G. stopped singing’ (= G. no longer sings). c. Gianni ha smesso di scrivere la tesi (= Gianni non scrive pit la tesi). ‘G, stopped writing his dissertation’ (= G. no longer writes his dissertation). d. %Gianni ha smesso di raggiungere la vetta (%#Gianni non raggiunge pit la vetta).??

‘G. stopped reaching the summit’ (G. no longer reaches the summit).

I will, thus, tentatively locate (non. . .) piitin the Spec of ASperminanves Teturming later to ASP perfect:

4.17 “Still” and continuative aspect In chapter 1, I provisionally assumed, following other work, that “still” is the positive counterpart of “no longer.’>° This might induce the conclusion that “still” occupies the same position as “no longer.” Since “no longer” seems related to “terminative”’ aspect, as just seen, and “still” to “continuative” aspect (see Haspelmath (1993, 145), cited in n. 2 of Appendix 2), we could take terminative and continuative to be two values (perhaps, marked and unmarked, respectively) of one and the same aspectual head.°! Some (slight) evidence exists, however, for taking (non. . .) pia and ancora as members of two distinct (though contiguous) adverb classes (and parallelly for keeping terminative aspect separate from continuative aspect). This is provided by the

fact that (on...) pitt and ancora can (marginally) co-occur, in this order, though not in the opposite one:°” (56) a. ?Spero che tu non sia pi# ancora arrabbiato con me! ‘T hope that you are no longer still angry with me.’ b. *Spero che tu non sia ancora piu arrabbiato con me! ‘T hope that you are still no longer angry with me.’

96 ADVERBS AND FUNCTIONAL HEADS

4.18 “Always” and perfect/imperfect aspect (?) Sempre ‘always’, when co-occurring with it, seems to follow ancora ‘still’ :>° (57) a. Gianni vince ancora sempre tutte le partite. ‘G. still always wins all the games.’ b. *?Gianni vince sempre ancora tutte le partite. ‘G. always still wins all the games.’ The projection hosting it must, then, follow AspPonunuative (and a fortiori AspP,erminatives TP(Anterior), Aspcejerative(t)» ASPP frequentative()> ASPP repetitive)» ASPPhabitual)-” In the hier-

archy of functional heads (and projections) discussed in chapter 3 (see (90)), the two heads immediately following T(Anterior) are Asppertect 20d ASPdurative: Sempre does not seem to enter a semantically transparent relation with Aspgyrative>

which refers to limited duration (“for a while”; see §4.19). Whether it should be related to ASP perfect/imperfect femains unclear. For concreteness, I will tentatively assume

it to relate to the imperfect value of this aspectual head, although the whole matter needs to be understood better. Another possibility is that it relates to what is sometimes called “continuous” aspect (or tense), to be kept separate from Continuative aspect (“keep on’/“still’”’). The New Guinea language Una, for example, has a suffix,

which Louwerse (1988, 30ff) refers to as “continuous tense,” translating it with ‘always’ (‘never’ when in co-occurrence with negation): (58) Er wa kum 1-ru-t ate, er tuba tuba bu-ru-t. (Louwerse 1988, 30) he garden not work-CONT T-3sg cc, he time time sit-CONT T-3sg ‘He never works in his garden, he always sits (on his rear end).’

Tepehua (a language of Eastern Mexico) is also reported to have a suffix (-(q) ‘ali-), which “signifies ‘always’” (Watters 1988, 244): milpa-‘ati-y “X always sings it’.

4.19 “Just”, “soon,” and retrospective and proximative aspects Various languages appear to have a form to express the fact that the event has taken place a short while before some reference time; a notion usually referred to in the literature as “retrospective aspect.’° Beyond the Haitian particle mentioned in chapter 3, n. 30 (see (i) of n. 55), aretrospective particle is found in other creoles and also in the Mayan language Mam (see England 1983, 162). In other languages, such as French and the Iberian Romance languages, the same

aspectual notion is expressed through a verbal periphrasis (venir de, acabar de; Comrie 1985, 94): (59) a. Je viens d’arriver. (French) ‘I come to arrive.’ ‘T’ve just arrived.’ b. Acabo de chegar. (Portuguese) ‘(1)} finish from arriving.’ ‘[’ ve just arrived.’

MATCHING AND REFINING THE HEIRARCHIES 97

In others, it is encoded as an affix on the verb. See the Yimas (Indo-Pacific) example

in (60a), from Foley (1991, 251), who refers to it as “immediate aspect,” and the Una (Indo-Pacific) example (60b), from Louwerse (1988, 63), who refers to it as “momentaneous aspect’: (60) a. Ti-n-ti-mpa-t. [Obj]-3sg[Subj]-do-IMM-PERF ‘She’s just finished.’

b. E-n-we. say-momAsp-I|sgiPAST ‘I was saying just a while ago.’

In other languages still, such as English or Italian, this notion of “recency” is ren-

dered through the use of adverbs like just/appena (Dahl 1985, 127), recently/ recentemente, lately/ultimamente, and so on, which in the present context it is natural to take as generated in the specifier position of this aspect projection.~® Just as there are (retrospective aspect) forms encoding the fact that an event took

place a short while before some reference time, so certain languages have forms encoding the fact that an event is going to take place a short while after some reference time. This aspect is sometimes referred to as “proximative” or “soon-aspect.” In Italian and English, adverbs semantically related to this aspect seem to be prestolsoon, subito/immediatamente/immediately, and others (Disse che avrebbe presto/subito lasciato I’Italia ‘He said he would have soon/immediately left Italy’ ).°’ Like appena (see n. 56), they also appear to necessarily follow sempre:>® (61) a. Disse che ci avrebbe sempre subito/immediatamente/?presto inviato sue notizie. ‘He said that he would have always immediately/soon sent news about him.’ b. *Disse che ci avrebbe subito/immediatamente/?presto sempre inviato sue notizie. ‘He said that he would have immediately/soon always sent news about him.’

The Eastern Malayo-Polynesian language Kwaio (Keesing 1985) also seems to point to the conclusion that retrospective and proximative aspects are two sides of the same coin, as the same particle appears to encode one or the other, depending on the tense of the sentence. “The particle bi’i following the subject-referencing pronoun is used both singly and in combination with the future marking particle. By itself it indicates recent completion of the action of the verb, and can usually be translated as ‘just’ . . . In conjunction with the future particle ta-, bi’i indicates that the action of the verb will take place at some (unspecified) time a short while in the future (“after a while” or “by-and-by”) (p. 119ff). Among others, he gives the following examples:°? (62) a. Ngai e bi”1 nigi. FPr(3s) SRP(1s) TAM arrive ‘He just got here.’ b. Ta-goru bi’i aga-si-a. FUT-SRP(1t) TAM see-TrS-Pro(3s) “We'll see 1t soon.’

It is not entirely clear, however, whether “retrospective” and “proximative” are two values of one and the same aspect head. As seen with “terminative” and “continua-

98 ADVERBS AND FUNCTIONAL HEADS

tive,” their appearing opposite, and complementary, dimensions might be a consequence of their semantics (and of their contiguity). In fact, just as terminative (non . . .) piu and continuative ancora were seen to co-occur in a fixed order, so are (certain combinations of) retrospective and proximative adverbs. See (63)—(64): (63) a. Gianni ha poco fa immediatamente accettato. ‘G, has a little ago immediately accepted.’ b. *Gianni ha ummediatamente poco fa accettato. ‘G. has immediately a little ago accepted.’

(64) a. Gianni ha recentemente subito interpellato il suo avvocato. ‘G. has recently immediately called his lawyer.’ b. *Gianni ha subito recentemente interpellato il suo avvocato. ‘G. has immediately recently called his lawyer.’

Given the lack of decisive evidence one way or the other, I will leave the question open, though favoring the two-aspect solution.

4.20 Durative adverbs and durative aspect The term “durative aspect” is generally applied in the literature to aspect markers that characterize a situation as “last[ing] for a certain period of time (or at least, . . . conceived as lasting for a certain period of time)” (Comrie 1976a, 41; see also Gibson 1992, 57). The plausible adverbial counterpart of this aspect is the class of durative adverbs like long, briefly, and others, and of durative (or measure; Moltmann 1991) adverbials like for an hour/a day/a month/and so on, for a while, and others.© The former, but not the latter, due to their phrasal nature, can appear in the speci-

fier position of a projection of the functional portion of the clause (see (65) versus * John has for an hour walked in the park). Adverbial PPs rather occur in the positions of circumstantial adverbials, within the VP. See §1.5. Whether they, or their features (Chomsky 1995), are able to move, and check the relevant features of the Spec of ASPgurative at LF, I leave open.

(65) Gianni ha brevemente/iungamente parlato delle sue vacanze. ‘G. has briefly/long been talking about his holidays.’

Here brevemente/lungamente mean ‘for a short/long time’. Relative to other adverb classes, significantly, they turn out to follow appena ‘just’ (and sempre ‘always’ ). See (66)—(67):

(66) a. Gianni ha appena brevemente parlato con il suo capo. ‘G, just briefly talked with his boss.’ b. *Gianni ha brevemente appena parlato con il suo capo. (irrelevantly possible with the nontemporal appena = ‘hardly’) (67) a, Gianni ha sempre lungamente parlato dei suoi problemi. ‘G. has always long been talking about his problems.’ b. *Gianni ha lungamente sempre parlato dei suoi problemi. ‘G. has long always been talking about his problems.’

MATCHING AND REFINING THE HEIRARCHIES 99

4.21 “2?” and generic/progressive aspect Generic sentences are sometimes treated together with habitual sentences. But the two

are plausibly to be kept separate. According to Dahl (1985, 97), habitual sentences “differ from generic ones by their lack of lawlikeness.” Generic sentences in fact seem to refer to some inherent characteristic (of an object) that may not even have had realization once. A sentence like Questa macchina fa i 280 km all’ora ‘this car runs 280 km per hour’ may be uttered appropriately even if the car has never been on the road. As a matter of fact, most of the times habitual adverbs are incompatible with a generic statement: %John usually is a man; *Whales are usually mammals.°! Here I will assume that generic sentences involve a generic operator in the Specifier position of an aspectual head which can also host (as its marked value) an “episodic” (Chierchia forthcoming), or “progressive,” operator. A well-known feature of the English present, as opposed to the present of Italian, and other languages is its compatibility only with the default (or generic) value of this aspectual head. The “episodic” value can only be realized via the progressive form. The incompatibility of individual level predicates with the progressive (*John is being a man; *Whales are being mam-

mals) can thus be attributed to their inherent genericity (Chierchia forthcoming). It is not clear what adverb class corresponds to this aspectual head (possible candidates are, in certain contexts, characteristically, inherently, typically, and so on).

4.22 “Almost/imminently” and prospective aspect As anticipated in n. 39 of chapter 3, the term “prospective aspect” has come to be used for those grammatical forms (affixes, particles, auxiliaries, and periphrastic constructions) that mark “a point just prior to the beginning of an event” (Frawley 1992, 322). This is for example the case with the English construction “to be going to’”’ / “to be about to” (Comrie 1976a, 64ff), or its equivalent in the Papuan language Kobon.® As clearly seen in Gungbe (§3.3 above), this aspectual head (immediately) follows the progressive head. It is sometimes analyzed as future tense, but such an analysis is not really warranted.®? See Comrie (1976a, 64). Adverbs that seem semantically related to this aspect are almost, nearly (in one of their meanings), and imminently, in English (and their analogues in other languages):™

(68) a. It was almost raining. b. He nearly accepted. c. We were imminently leaving for Spain.

The prospective periphrasis and “almost” show a comparable ambiguity with achievement predicates:

(69) a. Gianni stava per morire. ‘G. was about to die.’ b. Gianni é quasi morto. ‘G. has almost died.’ Both (69a) and (69b) can refer to a situation where a bullet just missed Gianni; or to a situation where he was hit by one, and this brought him near death.

100 ADVERBS AND FUNCTIONAL HEADS

4.23 Completely and tutto, and the two types of completive aspect In the hierarchy seen in chapter 1, completamente and tutto had been observed to follow sempre. As appena, subito, brevemente, and quasi also follow sempre (in that order), we may ask how these are ordered with respect to completamente and tutto. As (70)-(73) show, they precede completamente (and tutto): (70) a. Gianni ha appena completamente rovinato I’ arrosto, ‘G, has just completely ruined the roast-beef.’ b. *Gianni ha completamente appena rovinato l’arrosto. ‘G. has completely just ruined the roast-beef.’

(71) a. Lo avra tra poco/subito completamente rovinato. ‘He will have soon completely ruined it.’ b. *Lo avra completamente tra poco/subito rovinato. ‘He will have completely soon ruined it.’

(72) a. Loha brevemente completamente ridiscusso con Gianni. ‘He briefly completely discussed it over with G.’ b. *Lo ha completamente brevemente ridiscusso con Gianni. ‘He completely briefly discussed it over with G.’

(73) a. Gianni ha quasi completamente rinunciato alle sue pretese.© ‘G. has almost completely renounced his pretenses.’ b. *Gianni ha completamente quasi rinunciato alle sue pretese. ‘G. has completely almost renounced his pretenses.’

Concerning the location of completamente and tutto, I propose that they correspond to the two types of “completive” aspect discussed in the literature.°° Many languages appear to have a specific marker to signal that a telic process has reached comple-

tion (namely, the natural end point of the process). With a telic process like ‘eating the sandwich’, the natural end point is reached when the object has been totally affected (when there is no residue left of the sandwich). In English, this can be explicitly signaled with the particle up (He ate up his sandwich, Eat up your sandwich/); in Hungarian with the (separable) prefix meg- (Eda meg a szendvicsedet ‘Eat up you sandwich’). In other languages, it is signaled by a (completive aspect) suffix. See: (74) a. Wara-kaaku-sha. (Huallaga Quechua, Weber 1989, 152) dawn-COMPL-3PERF ‘It has completely dawned (..e., it is now day).’ b. Ta zuotian xie-wan-le yifeng xin. (Chinese, Smith 1987, 96) he yesterday write-COMPL-PEREF a letter ‘He wrote a letter (to the end).’ c. Kuugal timm-id-1. (Fula/Fulfulde, Fagerli 1994, 19) work finish-COMPL-TENSE “The job is completely finished.’

In others, it is expressed by an auxiliary. See the Choctaw (Muskogean, North America) case discussed in Schiitze (1995, 449) and the Ainu (Altaic) case in n. 68.

MATCHING AND REFINING THE HEIRARCHIES 101

In the case of a plural (definite) object, “completion” implies both of two things: (1) that the plural set has been totally affected (i.e., each member of the set has been affected), and (2) that each member of the set has been totally affected (see Bybee

et al. 1994, 57).° In many languages, these two senses of “completion” are not formally distinguished, nor can they be isolated one from the other. In English, for example, J ate up the sandwiches implies at the same time that all the sandwiches were affected and that each was affected completely. It would be inappropriate to utter such a sentence if only one bite was taken from each sandwich, or if only two thirds of the sandwiches were eaten. But there are languages that keep these two senses separate, apparently. According to Siewierska’s (1991, 122) characterization, Polish has two completive prefixes (which she refers to as “completive” and “perfective”), one expressing completion of the set (“one after the other’), the other expressing completion of each item of the set (“right through”):®8 (75) Po-prze-czyt-yw-o am wszystkie je} kSiazki. COMPL-COMPL-read-HAB-PAST all her books ‘I have read all her books occasionally one after the other and right through.’

For ease of reference, I will call the first “plural completion” and the second “singular completion.” It is tempting to see completamente and tutto as the specifiers corresponding to “singular” and “plural” completion, respectively. For bare tutto 1mplies a plurality of items. One could not appropriately utter (76) if he knew that just one item was been looked for:

(76) Hai trovato tutto? ‘Have you found everything?’ The specifiers and the completive aspect suffixes and particles are characterized by a common restriction. They are all incompatible with situations lacking internal stages and a natural end point. See the ill-formedness of the examples in (77) (involving activities), from Hungarian, Chinese, and Italian, respectively: (77) a. *Karoly tejet ivott meg. (de Groot 1984, 138) K. milk drank up ‘K. drank milk up.’ b. Ta xiao-(*wan)-le. (Smith 1987, 119) he smile-(COMPL)-PERF ‘He smiled completely.’ c. *Gianni ha riso completamente. ‘G. smiled completely.’

As noted in chapter 1 (n. 57) completamente can, in fact, occupy two distinct positions; a preverbal and a postobject one, associated with two distinct interpretations, which likely depend on their different scope. In §4.26, we will see some reasons to postulate an additional completive aspect projection.

4.24 “Well” (manner adverbs) and voice As seen in chapter 1, the adverb class next to tutto appears to be that of bene (and other light-manner adverbs). A derivational suffix meaning “well done” is appar-

102 ADVERBS AND FUNCTIONAL HEADS

ently found in Chadic languages (Newman 1992, 254). Here I tentatively suggest that this class of adverbs occupies the specifier position of VoiceP, even though the relation between the two is not self-evident. Clues in this direction are the close link existing between “Middle Voice” and manner adverbs (see Keyser and Roeper 1984, 384; Roberts 1986, 194ff), and the special morphological relation existing in certain languages between manner adverbs and Passive Voice. In Maori (Austronesian), for example, “the manner particles passivize in agreement with passive verbs” (Bauer 1993, 92). See (78): (78) I peehi a rawa tia ngaa waahine. T/A oppress PASS intens PASS the(pl) women ‘The women were severely oppressed.’

The existence of a special relation between Passive and manner adverbs is also suggested in Chomsky (1965): “The Verbs that do not take Manner Adverbials freely Lees has called “middle Verbs’ (Lees 1960, 8), and he has also observed that these are, characteristically, the Verbs with following NPs that do not undergo the passive transformation’ (p. 103) [the verbs in question are resemble, have, marry, fit, cost, weigh, etc., in one of their senses]. ““The generalization that relates Manner Adverbials

to passivization” (Chomsky 1965, 218, n. 28) is formally represented in Aspects in the rewriting rule: Manner — by passive. There is also a word order peculiarity in Romance that can be interpreted as suggestive of a relation between Passive and manner adverbs. As observed in chapter 2, an active past participle in Italian necessarily precedes bene (and tutto). See (79): (79) a. Hanno accolto bene il suo spettacolo solo loro. have received well his show only they b. *Hanno bene accolto il suo spettacolo solo loro. have well received his show only they

Passive past participles, however, seem to behave differently. At least under certain conditions, to which I return, they can also be found to the right of bene ‘well’. See (80): (80) Questo genere di spettacoli é sempre stato bene accolto da tutti. ‘This kind of show has always been well received by everybody.’

A possible reason why (79b) is ill-formed is the fact that, in Italian, active past participles raise in overt syntax to check the feature “perfect” in the relevant head, which is higher than Aspcompietive and Voice. If bene is in Spec, VoiceP, then the active past

participle necessarily precedes bene. Things are different with passive past participles, however. In this case, it is to be expected that overt raising can be limited to Voice’, where (the marked value) “passive” has to be checked. If bene is in Spec, VoiceP, then we expect it to be found to the left of the participle. This is correct, as noted, though only in part, as the other order is also possible: (81) Questo genere di spettacoli é sempre stato accolto bene da tutti. ‘This kind of show has always been received well by everybody.’

This is not areal problem. It is still possible that, mn (81), the passive past participle, beyond checking the marked feature of Voice® (its primary function), has the fac-

MATCHING AND REFINING THE HEIRARCHIES 103

ulty of raising and checking some other (marked) feature in a higher head. This seems to be generally the case with active past participles in Italian, which in addition to checking the (marked) feature “perfect” (perhaps their primary function) are apparently able to check the marked feature E_R, of T(Anterior) (whence their compatibility with such adverbs as ieri ‘yesterday’, and others). A clue to what this additional head might be, in the case of Italian passive past participles, comes from a special restriction holding of the order ‘bene + passive past participle’. This order is only acceptable in generic sentences such as (80). If a specific time reference is involved, the order becomes impossible. See (82): (82) a. *Ieri sera, il suo spettacolo era stato bene accolto da tutti. ‘Last night, his show had been well received by everybody.’ b. Ieri sera, il suo spettacolo era stato accolto bene da tutti. ‘Last night, his show had been received well by everybody.’

I interpret this as indicating that when a specific time reference is involved, the passive past participle has to raise and check the marked feature of the generic/progressive head (which is “nongeneric,” or “progressive’’), in addition to the marked fea-

ture of Voice” (“passive”). Only when the higher head has the default feature (“generic”) (so that no checking is required), can the passive past participle fail to raise, and remain in Voice’.”!

4.25 “Quickly/fast/early” and celerative aspect (Il) As noted in §4.14 (after Travis (1988)), adverbs of the class of quickly can appear in two different positions, with partially different interpretations. The higher (the one quantifying over the event) occurs in between frequentative (1) adverbs and gid ‘already’ (see (48)-(5Q0)). The lower (the one quantifying over the process) appears instead to occur in between bene ‘well’ and (the lower) completamente ‘completely’ (for which see the next section): (83) a. I candidati lo tradussero bene presto tutti.” ‘The candidates translated it well quickly/early all.’ b. *I candidati lo tradussero presto bene tutti. ‘The candidates translated quickly/early well all.’

(84) a. Dimenticherai tutto presto completamente anche tu. “You will forget everything quickly completely too.’ b. *Dimenticherai tutto completamente presto anche tu. “You will forget everything completely quickly too.’

In English, while quickly, rapidly, and others can occur in either position, the synonymous adverb fast can occur only in the lower of the two positions (being apparently unable to quantify over events):” (85) a. He quickly ran away. b. He ran away quickly.

(86) a. *He fast ran away. b. He ran away fast.

104 ADVERBS AND FUNCTIONAL HEADS

This lower position (of guickly and fast) seems to correspond to the specifier of the celerative aspect head of Fula/Fulfulde (Arnott 1970, 356; Fagerli 1994, 36ff), which is realized as a (derivational) suffix to the left of the Voice suffix (which belongs to an inflectional complex also expressing aspect and polarity), and to the right of the Completive aspect (derivational) suffix (Fagerli 1994, 53). This matches precisely (in a mirror fashion) the order of what I take to be the corresponding specifiers bene, presto, and completamente, in Italian. See (83)-(84). Arnott (1970), in the passage quoted in n. 74, points out that the celerative aspect suffix can either mean “quickly” or “early.” Interestingly, the adverb presto, in Italian, is also ambiguous between these two meanings (see the preceding translation of (83a)). Like fast (and, for that matter, early) in English, presto cannot occupy the Spec of the higher Asp pejerative. 2€€ (87) Versus (83a), and (88a—b):

(87) Gianni ha rapidamente (*presto) risolto i suoi problemi. ‘G. has quickly/early solved his problems.’

(88) a. Today, I woke up early. b. *Today, I early woke up.

4.26 “Completely” and completive aspect (II) As noted in chapter 1, n. 57, completely can occur in two distinct positions, with partially different interpretations. The higher position was seen, in §4.23, to be located between quasi ‘almost’ and (the derived position of) tutto ‘everything’. The lower

position appears instead to come, as just noted, after bene ‘well’ and presto (rapidamente/velocemente/etc.) ‘quickly’. In terms of order, it seems to correspond to the specifier position of the Completive aspect head of Fula/Fulfulde, realized with the (derivational) suffix -(i)d. Just as the order of suffixes in Fula/Fulfulde is (V)COMPLETIVE-CELERATIVE-VOICE, the order of specifiers, in Italian, 1s reversed (consistently with the Mirror Principle): bene > presto > completamente.

4.27 Repetitive/frequentative adverbs and repetitive/frequentative aspect (Il) In Fula/Fulfulde, in between the Completive aspect suffix and the verb it is possible to have a Repetitive aspect suffix, meaning “again” (see Fagerli 1994, 41, 53). (Also see the case of Yavapai (Hokan, North America), where the repetitive suffix (again, once more”) occurs between the verb and the progressive suffix: mi:-yi-km °X 1s crying

again’; Kendall 1976, 30.) This might suggest locating the lower Repetitive and Frequentative aspect projections, mentioned in $4.13, below the Completive aspect projection. But the order of the corresponding adverbs of Italian would seem to suggest that they are higher than Asp.ompletive (ID: (89) a. Ha gia dimenticato di nuovo piu volte completamente 1 nostri compleanni. ‘He already forgot again several times completely our birthdays.’

I leave the question open, noting a potential problem.

MATCHING AND REFINING THE HEIRARCHIES 105

4.28 Speculative remarks on other aspects and adverb classes In this section, I briefly mention other possible (aspectual) functional heads and adverb classes, whose nature and location remains to be precisely determined. Part of the reason for listing them here is to emphasize the incomplete nature of the analysis presented in this study. The assumption, of course, is that the number of such projections is finite and, plausibly, quite limited, and that it should eventually be possible to arrive at fully determining them on empirical grounds. Various languages mark morphologically (e.g., through a verbal suffix) the beginning of a process (the so-called inceptive, or ingressive, aspect). See the suffix-kri of Boro mentioned in n. 79, or the case of Aleut and Ika in Appendix 2 (exs. (16a) and (40b)).’° Other languages mark the fact that a certain action may require some effort (the so-called “conative” aspect), or that “a distance is traversed before the action is done,” as with the “prefix d’éloignement” of Tinen (Dugast 1971, 170), and the “distantive” (or “andative”’) suffix of Fula/Fulfulde (Fagerli 1994, 35). See (90) from Fula/Fulfulde:”” (90) Mi rem-oy-i ngesa. I cultivate-DIST-PAST field ‘I went and cultivated the field.’

A verbal morpheme indicating “that an accomplishment takes place or succeeds only

through extra effort” (Carlson 1996, 59) is documented in the Salishan (North America) language Spokane. Carlson refers to it as the “SUCCESS” morpheme, translating it with manage, or the adverb finally. See also the Mongo (Bantu) affix -ey-,

which Larochette (1980, 34) calls “capacitatif,” and renders with ‘réussir’ (‘manage’) (-kol—‘faire’, -koley- ‘réussir a faire quelque chose de difficile’). Possibly related to this aspect are the “frustrative” morphemes found in Wayampi

(Tupi-Guarani; South America) and other languages, which Jensen (1994, 359ff) renders as “without success.” See also the verbal suffix of the Cariban (South America)

language Macushi in (41d) in Appendix 2, rendered as “finally” by Abbott (1991), and the “delayed” aspect particle of Ulithian (Austronesian), glossed by Sohn and Bender (1973, 116) as ‘finally.’ Another aspect discussed in the literature is the so-called delimitative (Comrie 1976a, 22, n. 3), or pofective (Pifion 1993) aspect, which “indicate[s] a temporally restricted, but non punctual situation.” While it is clear to what adverbials it corresponds (“for a little time,” etc.), it is not clear whether it is distinct from Durative aspect (see §4.20). Tuyuca (Tukanoan, South America) is reported in Barnes (1994, 331) to have an aspect suffix which she glosses as “constantly” (a relevant example is given in chapter 5 in this volume, n. 51). The existence in Pawnee (Caddoan, North America) of a “usitative” aspect preverb, meaning ‘occasionally’ (distinct from the habitual aspect suffix), was mentioned here in n. 37. It is unclear, however, whether it is to be kept distinct from frequentative aspect. A “pretense” (or “pretensive’’) aspect meaning “pretend to, act as if” 1s attested in various languages. See, for example, the case of Coahuilteco mentioned in Ap-

106 ADVERBS AND FUNCTIONAL HEADS

pendix 2, or that of Fula/Fulfulde, in which evidence exists for its location below Voice (see Fagerli 1994, 38, who calls it “simulative’’). “Pretense” prefixes are also found in the Austronesian language Da’a (Barr 1988, 36), in the Caucasian language Evenki (Nedyalkov 1997, 270), and in the Mon-Khmer language Pac6h (Watson 1966, 29). See also the suffix -ké- of Boro (Sino-Tibetan), cited in n. 79, meaning ‘pretend to’. Other languages have suffixes sometimes referred to as “priorative’” or “sequential,” which mark an event as occurring before another. In Tepehua (Totonacan, Mexico), “the suffix - ‘el means roughly ‘to do first (before something else)’ (Watters 1988, 243): (91) Pas-‘ela-t es‘a-i la: “18caga:. bathe-SEQ-PERF then go-PERF PREP 3Poss-house ‘*X bathed first, then went to his/her house.’

As Watters further notes (1988, 243), “this suffix is thus semantically redundant with

the adverb p’unah ‘first’, with which it often occurs, or which often occurs in its stead.”’®

4.29 Toward a universal hierarchy of clausal functional projections (a second approximation) The preceding discussion was meant to offer evidence that the hierarchies of adverbial specifiers and clausal functional heads match in a systematic one-to-one fashion. Even if we abstract from the more tentative functional heads and adverb classes of the preceding section, whose integration with the rest requires further study, the functional structure of the clause that we arrive at (see (92)) is very rich; at first sight, outrageously rich: (92) The universal hierarchy of clausal functional projections (a second approximation) [ frankly Mood gpecch act | fortunately Moodgyatative | allegedly Mood ey dential

[ probably Modaenisteme [ once T(Past) [ then T(Future) [ perhaps Moodireats | necessarily Modhecessity | possibly Modpossinty | usually Asppabiual | again ASP yepetttive(!) | often ASP frequentative(1) | intentionally Mod, oiitional

[ quickly ASP ceteravea) | already T(Anterior) [ no longer ASPyemnanve | Still ASP ontinuative [ always ASPperfect( | just ASP retrospective | soon ASPmoximative [ briefly ASPauratve [ Characteristically(?) ASPgenericiprogressive | @MOSt ASPprospective

[ completely Aspsecompetivecy | tutto ASPpicompienve | well Voice [fastlearly ASD coteranvec | Again ASPrepetitrve(d) | often ASPfrequentauvec) | completely ASPgocompleuvetI)

Viewed from the specifiers’ side, this richness should not appear so outrageous, though. Languages are generally much richer in the realization of different classes of AdvPs than in the realization of the corresponding heads (through affixes, particles, auxiliaries, etc.), although there are exceptions.” In this connection, if each adverb class indeed corresponds to a different functional head, then, we have evidence that the entire array of functional heads (and projections) is available even where there is no overt morphology corresponding to the heads, as the respective specifiers are available.

MATCHING AND REFINING THE HEIRARCHIES 107

Suppose, however, that one were to reject this conclusion, granting the existence of a certain functional projection only in the presence of overt morphological material on a head. Since, in this case, (most) adverbs would not be systematically related to a functional head, UG would have to countenance two distinct conditions (one ruling over the hierarchy of heads, the other over the hierarchy of AdvPs), basically yielding (duplicating) the same information on the relative scope of what are essentially identical functional notions. An undesirable consequence, it seems.*° Also, no particular acquisitional issue arises from considering the structure of the clause provided by UG as rich and articulated as that in (92). The obvious consequence from assuming the universality of (92) is that less is left for the child to acquire. He/she will only need to recognize and locate in the appropriate structural places

made available by UG the morphological and lexical material provided by his/her language. It would hardly be possible for an Italian child to determine on empirical grounds what the relative position is of, say, (won. . .) piu “no longer’ and ancora ‘still’ (marginal in one order, and impossible in the other; see §4.17). Their relative order should rather follow from UG once the meaning of each adverb/specifier is recognized. The same should hold for the relative order of two suffixes, or particles, in the case of heads.

5

DP-Related Functional Projections and Negative Phrases

So far, nothing has been said about the distribution of DPs and of negation. The enlarged functional structure of the clause suggested here, however, has direct implications regarding their positions. Consider first DPs. If AdvPs occur in the specifier of distinct, rigidly ordered, functional projections, the fact that DPs (and floating quantifiers, as a special case) can be found interspersed among them implies the existence of several DP-related positions (many more than is usually assumed). This, in turn, raises the question of whether such positions are themselves specifier positions, adjunction positions (or a combination of the two). The evidence to be discussed points to their being specifier positions of separate (DP-related) functional projections; a conclusion also forced on theoretical grounds in Kayne’s (1994) system. To call them Agreement Phrases is perhaps not very illuminating. As Chomsky (1995, chap. 4) notes, if all there is to agreement is a morphological relation (with no LF relevance) between a DP in specifier position and the corresponding head, little justification remains for positing the existence of an independent (AgrP) projection (see also Mitchell 1993). The morphological agreement between.a DP and a head could be the overt reflex of more fundamental abstract notions—a way to overtly mark the head of (certain) DP-related projections. Recent work on the scope properties of different DPtypes (Beghelli 1995; Beghelli and Stowell 1997), or on the different interpretations associated with different “scrambling” positions (Moltmann 1990; Diesing 1992, 1996a,b; Diesing and Jelinek 1995; Haiden 1996, among others), makes it tempting to take (some) such DP-related projections to be positions “specialized” for particular readings (e.g., “existential,” “distributive,” or “specific”), or particular scopes. This is in line with the current conception of grammatical relations, according to which, as McCloskey (1997, 216) phrases it, “there is no ‘subject position’—1in the 108

DP-RELATED FUNCTIONAL PROJECTIONS AND NEGATIVE PHRASES 109

sense of a unitary position in which all subject properties are expressed and licensed. Rather, subject properties are distributed over a sequence of derivationally linked positions” (a theta-related one, a Case related one, and so on). Conceivably, in addition to the distinct “subject” (or “object’’) positions related to purely grammatical functions, there could be others with interpretive functions. I will refer to them as “DP-related projections,’ though ultimately a more perspicuous terminology should be adopted. Here, I will be able to offer only rather speculative remarks. The facts and the generalizations involved have just begun to be systematically investigated.!

The evidence for assuming different DP-related functional projections interspersed among the Adverb-related functional projections comes from the distribution of subject (§5.1) and object (§5.2) DPs, from the distribution of floating quantifiers (FQs) under the stranding analysis of Sportiche (1988), ($5.3), and, crucially, from the distribution of the verb. If a verb can always be found to the immediate right of a DP (or FQ), a direct argument exists for taking the latter to be in the specifier position of an independent, full-fledged, functional projection rather than adjoined to (or in a “second” specifier of) the projection hosting the AdvP that follows. Concretely, the two approaches lead to quite different expectations in the case of sequences like AdvP, DP AdvP, (or AdvP, FQ AdvP,). In the approach in which DPs (and FQs) are adjoined to (or are a “second” specifier of) the projection hosting AdvP,, there is no head position between the DP (or FQ) and AdvP,. Consequently, no verb should occur there.’ In the approach in which DPs (and FQs) are necessarily specifiers of an independent functional projection found between those hosting AdvP, and AdvP,, we instead expect a verb to be able to occur, as there is a head position available there. That is apparently what we find (in Italian). See §5.1.° A related general issue is the following. Given the evidence that DPs, and verbs, move within the sentence, the question arises whether such alternations as (la—c) can be exhaustively accounted for by DP- and V-movement, or whether one needs to assume movement, or multiple “base generation,” of AdvPs as well: (1) a. Probably George will have read the book. b. George probably will have read the book. c. George will probably have read the book. (2) *George will have probably read the book.*

The discussion in Chomsky (1995, 48; 390, n. 102), and in $1.4 here, suggests on various grounds that AdvPs do not move (except to specified operator positions, or in the special “adverb climbing” construction of French seen in $1.4). If, furthermore, each AdvP is necessarily generated in the unique specifier position of a related functional head, such alternations as (1a) and (1c) cannot involve multiple “base generations” (or adjunctions) of the AdvP, either. They can only involve the movement of the subject DP and of the verb to different positions “around the AdvP.” This is the course I take and explore in this study.° A partly similar situation characterizes (sentence) negation, which is also found to occur in more than one position along the hierarchy of AdvPs, sometimes simultaneously. For a discussion on the positions of NegPs within the enlarged functional structure of the clause, see $5.4, and the references cited there.

110 ADVERBS AND FUNCTIONAL HEADS

5.1 The positions of subject DPs One first piece of evidence for the presence of different DP-related functional projections is provided, as noted, by the variable positioning of the subject DP, and the verb, along the rigidly ordered sequence of AdvPs in the sentence. In fact, the postulation of a universal hierarchy of adverb-related projections like the one proposed in §1.7 and refined in chapter 4 leads to certain expectations. So, for example, if a (finite) verb can appear to the left of a certain adverb class, then it should necessarily be able to appear to the left of all adverbs lower than that in the hierarchy (though not necessarily to the left of any higher one). Also, if a verb cannot appear to the left of a certain adverb class, a fortiori, it should not be able to appear to the left of any higher one (see chapter 1). Similar predictions can be formulated in relation to the positions of (subject and object) DPs with respect to various adverb classes. As we are going to see, just as they differ as to the positions in which the various verbal forms are found (see Appendix 1), some Romance and Germanic languages also differ as to the positions they make available to subject and object DPs along the fixed hierarchy of AdvPs. To begin, consider the position of the subject in Italian. This must precede mica ‘not’, and, consistently, all adverbs lower than mica:®

(3) a. Maria mica prende il treno. *M. not takes the train.’ b. *Mica Maria prende il treno. “Not M. takes the train.’ c. *Gia Maria é di ritorno, per le una. ‘Already M. is back, at one o’clock.’ d. *Pit Maria non mi pensa. ‘No longer M. thinks of me.’ e. *Ancora Maria gli parla. ‘Still M. speaks to him.’ f. *Sempre Maria ripete le stesse cose. ‘Always M. repeats the same things.’ g. *Appena Maria si era coricata, quando squilld il telefono. ‘Just M. had gone to bed, when the phone rang.’ h. *Subito Maria mi avvertiva. (no focus intonation on subito) ‘Immediately M. would call me.’ i. *?Brevemente Maria ci sta parlando della sua avventura. ‘Briefly M. is telling us about her adventure.’ 1. *Quasi Maria cadde dall’emozione. ‘Almost M. fell for the emotion.’ m. *Completamente Maria distrusse tutto quello che aveva fatto fino ad allora. ‘Completely M. destroyed all that she had done till then.’ n. *Bene Maria fece tutti 1 compiti. “Well M. did her homework.’ o. *Presto Maria si alzava ogni mattina. ‘Early M. would get up every morning.’

The subject, on the other hand, can either precede or follow all adverbs higher than mica (with no gaps):’ (4) a'. Rapidamente Gianni alzo di nuovo il braccio. ‘Quickly G. raised again his arm.’ a", Gianni rapidamente alzo di nuovo il braccio. ‘G. quickly raised again his arm.’ b'. Raramente Gianni rifa tutto bene. ‘Rarely G. does everything well again.’ b". Gianni raramente rifa tutto bene. ‘G. rarely does everything well again.’ c'. Di nuovo Gianni rifiuto |’invito. ‘Again G. refused the invitation.’ ec". Gianni di nuovo rifiuto |’ invito. ‘G. again refused the invitation.’ d’. Solitamente Gianni pranza alle due. ‘Usually G. eats at two.’ d". Gianni solitamente pranza alle due. ‘G. usually eats at two.’ e'. Stupidamente Gianni accetto di venire. ‘Stupidly G. accepted to come.’

DP-RELATED FUNCTIONAL PROJECTIONS AND NEGATIVE PHRASES 111

e". Gianni stupidamente accetto di venire. ‘G. stupidly accepted to come.’ f'. Obbligatoriamente le lezioni iniziavano il primo ottobre. ‘Obligatorily classes began the ist of October.’ f". Le lezioni obbligatoriamente cominciavano il primo ottobre. ‘Classes obligatorily began the Ist of October.’

¢'. Volentieri Gianni si offri di aiutarci. “Willingly G. volunteered to help us.’ g", Gianni volentieri si offri di atutarci. ‘G. willingly volunteered to help us.’ h'. Non necessariamente 1 preti sono pacifist. ‘Not necessarily priests are pacifists.’

h". Ipreti non necessariamente sono pacifisti. ‘Priests not necessarily are pacifists.’

i. Forse Gianni verra a chiamarci. ‘Perhaps G. will come and call us.’ i". Gianni forse verra a chiamarci. ‘G. perhaps will come and call us.’ j. Allora Gianni era monarchico. ‘Then G. was for the monarchy.’ jy". Gianni allora era monarchico. ‘G. then was for the monarchy.’ k'. Probabilmente Gianni ha rinunciato. ‘Probably G. has renounced.’ k". Gianni probabilmente ha rinunciato. ‘G. probably has renounced.’ I’. Evidentemente Gianni é contento cosi. ‘Evidently G. is happy with that.’ I". Gianni evidentemente é contento cosi. ‘G. evidently is happy with that.’ m'. Purtroppo Gianni ha accettato. ‘Unfortunately G. has accepted.’ m", Gianni purtroppo ha accettato. ‘G. unfortunately has accepted.’ n'. Francamente Gianni ha esagerato. ‘Frankly G. has exaggerated.’ n'. Gianni francamente ha esagerato. ‘G. frankly has exaggerated.’ Note that the subject can either precede or follow any adverb higher than mica, provided that the (finite) verb follows it (as in (4)).® The only time the (finite) verb is allowed to precede the subject in Italian is when the subject is in the “inverted,” sentence final, position (see (Sa)). Any other postverbal, non absolute final, position gives rise to ungrammaticality (under normal intonational conditions). See (Sb-—c): (5) a. Hanno comprato il giornale i Rossi. (see Burzio 1981, chap. 2, n. 5) ‘Have bought the newpaper the Rossi’s.’ b. *Hanno 1 Rossi comprato il giornale. c. *Hanno comprato i Rossi il giornale. (possible only with il giornale ‘deaccented’ and prominence on i Rossi)

I tentatively interpret this situation as follows. The sentence final position coincides with the Spec of VP (Kayne 1994, Ordéfiez 1994), all higher positions involving the Spec of a DP-related functional projection. Suppose that an expletive pro can fill one of the DP-related functional projections only when the subject remains in VP (this can be enforced by assuming that expletive pro 1s necessarily generated in the Spec of the lowest DP-related functional projection, later raising to a higher one). As a consequence of that, whenever the subject fills the Spec of one of the DP-related functional projections outside VP (via the lowest one), we know that expletive pro could not possibly have been generated. Suppose, furthermore, that the agreeing finite verb must be (at the Spell-out point) in a Spec/head relation with a DP subject (or its trace). This entails that the subject either remains in Spec VP (“doubled” by the expletive, which enters the required Spec/ head relation with the verb), or, whenever it is outside the VP, it necessarily comes to precede the finite verb (as no expletive is available to satisfy the Spec/head requirement).

112 ADVERBS AND FUNCTIONAL HEADS

If 1t remains in Spec VP, one may wonder why the subject cannot precede the complements (see the ungrammaticality of (5c)). As a matter of fact, it can, provided it is a nonspecific indefinite (see Beninca’ 1988, 124): Se per allora avra comprato qualcuno il giornale, lo sapremo anche noi ‘If by then will have bought someone the newspaper, we’ll know it too’. Apparently, a definite subject can remain in Spec VP only if it bears narrow (contrastive) focus, which implies (in Italian) that everything following it must cross over it (see Cinque 1993).? A similar situation holds in Icelandic. While a definite subject has to precede all adverbs, an indefinite or a contrasted definite subject (Vangsnes, forthcoming,

and references cited there) has to remain (preferably remains) lower than alltaf ‘always’, though higher than reet ‘just’ and adverbs lower than that (Sigurdsson, personal communication). As 1n Italian, the subject in English has to precede already (English has no analogue of mica), and all adverbs lower than already: (6) a. Frequently John takes his holidays abroad. b. Quickly John raised his arm. c. *Already John knows that you are coming. d. *No longer John likes Mary. e. *Still John misses Mary. f. *Always John takes his holidays abroad. g. *Just John has left. h. ?Soon the train leaves. So hurry up!!° i. *?Briefly I consider it in my book. 1. *Almost John fell through fright. m. *Completely John destroyed all that he had constructed. n. *Well John did his homework. o. *Early John woke up every Sunday.

Again as in Italian, the subject may either precede or follow any higher adverb, provided it precedes the verb (see the translations of the Italian sentences in (4)). Italian provides direct evidence that a head position exists between the subject and the following adverb in all of the “double prime” examples in (4) (hence that the subject comes to occupy the specifier position of a distinct functional projection). The evidence is that the finite verb can be found in each such position, with no exceptions (Gianni ha rapidamente alzato di nuovo il braccio ‘G. has quickly raised again his arm’; Gianni ha raramente rifatto tutto bene ‘G. has rarely done everything well again’; etc.). Although the same cannot be shown for all languages (see the case of Norwegian in (7)), due to whatever forces the verb either raise to C, or remain in a low position (in non-V/2 contexts), I will assume, for generality, that the same head positions are available there, as they are in Italian. This conclusion is in fact forced in a system like that of Kayne (1994). (7) a. Na skj@nner tydeligvis allerede ikke lenger alltid Per alt helt godt. ‘Now understands evidently already no longer always P. everything completely well.’

b. Na skj@nner tydeligvis allerede ikke lenger Per alltid alt helt godt. c. Na skjgnner tydeligvis allerede Per ikke lenger alltid alt helt godt. d. Na skj@nner tydeligvis Per allerede ikke lenger alltid alt helt godt. e. Na skjgnner Per tydeligvis allerede ikke lenger alltid alt helt godt.

DP-RELATED FUNCTIONAL PROJECTIONS AND NEGATIVE PHRASES 113

f. ... at snart Per leser den boka i filler. ‘,.. that soon P. reads that book to pieces.’ g. ...atnettopp Per forlot festen i raseri. *... that just P. left the party in rage.’ h. *...at kort Per forklarte problemet. *,.. that briefly P. explained the problem.’ i. *,.. at nesten Per leste den boka 1 filler. *,,. that almost P. read that book to pieces.’ l. *... at tidlig Per forlot den festen i vildt raseri. ‘,,. that early P. left that party in wild rage.’

In Norwegian, the subject has to precede kort ‘briefly’, nesten ‘almost’, helt “completely’, godt ‘well’, tidlig ‘early’, but it can (apparently optionally) be found to the right, or left, of any higher adverb (recall, from §1.7, the (partial) order of Norwegian adverbs: erlig talt ‘honestly’ > heldigvis ‘fortunately’ > tydeligvis ‘evidently’ > kanskje ‘perhaps’ > klokelig ‘wisely’ > allerede ‘already’ > ikke lenger ‘no longer’ > alltid ‘always’ > helt ‘completely’ > godt ‘well’).'! From the position to the left of kort leftward, there are no gaps in the possibility for the subject to occur between any two adverbs (and to the left of the highest).!* The synopses in (8a, 8b, and 8c) summarize the main facts concerning the distribution of the subject in Italian, English, and Norwegian (here Italian and English pattern together, contrasting with Norwegian): (8) a. V onestam V fortunatam V evidentem V probabilm V ora V forse V b. V honestly V luckily V evidently V probably V now V perhaps V c. V erlig talt V heldigvis V tydeligvis V sannsynligvis Vv na vV kanskje V a. V necessariam Y¥ volentieri Vobbligatoriam Vsaggiam V di solito ¥ di nuovo V b. V necessarily V willingly V obligatorily Vv wisely V usually V again V c. V ngdvendigvis V gjerne Vobligatorisk ¥ klokelig V¥ vanligvis V igjen V

a. V spesso V rapidam V gia *(non) pid * ancora * sempre * appena * subito * b. V often V quickly V already * no longer * still * always * just * soon * c. V ofte V raskt V allerede V ikke lenger V enda V alltid ¥ nettopp V snart V

a. * brevemente * quasi * completam * bene * presto* b. * briefly * almost * completely * well * early * c. V kort * nesten * helt * godt * tidlig *

More difficult questions are those pertaining to (1) the nature of these various subject positions, (2) the obligatory movement of the subject to a position to the left of a certain (“lower’’) adverb (mica in Italian, already in English, and kort in Norwegian),!° and (3) the apparently optional movement of the subject around the adverbs higher than the position of obligatory movement. Concerning the first question, it appears that some such positions are interpretively specialized, as noted. It is often remarked in the literature that the same DPtype admits one or the other of two different interpretations depending on its position with respect to certain adverbs. Diesing (1992, 36ff; 78ff) observes, for example, that bare plural subjects in German receive an existential reading when appearing to the right of such adverbs as ja doch, and a generic one when appearing to their left (see (9)); subjects with

114 ADVERBS AND FUNCTIONAL HEADS

numeral determiners receive in the two positions a presuppositional, and a cardinal, interpretation, respectively (see (10)):'4 (9) a. ... weil ja doch Haifische sichtbar sind. since ‘indeed’ sharks visible are ‘... since there are sharks visible.’

b. ... weil Haifische ja doch sichtbar sind. since sharks ‘indeed’ visible are ‘,.. Since (in general) sharks are visible.’

(10) a. ... weil ja doch zwei Cellisten in diesem Hotel abgestiegen sind. ‘,,. Since ‘indeed’ two cellists in this hotel have taken rooms.’ b. ... weil zwei Cellisten ja doch in diesem Hotel abgestiegen sind. ‘.,. since two cellists “indeed’ in this hotel have taken rooms.’

Diesing (1992) takes the two positions demarcated by ja doch to be Spec, IP and Spec, VP, respectively; but finer-grained distinctions are required if more adverb classes, and DP positions, exist.!> Beghelli (1995) and Beghelli and Stowell (1997, 96) propose a hierarchy of distinct DP/QP-related functional projections, to the Spec of which different DP-types raise (at LF). See (11). In this way, they are able, among other things, to reduce the number of scope ambiguities that would be expected under an undifferentiated QR derivation of quantificational DPs.

(11) Ref(erential)P Group QPs CP Wh-QPs Agr SP

Count QPs JN Distributive QPs ShareP Group QPs Neg(ative)P Negative QPs Agr ioP Count QPs AgIpoP

Count QPs VP

DP-RELATED FUNCTIONAL PROJECTIONS AND NEGATIVE PHRASES 115

What remains to be done is see where these projections are located relative to the various adverb-related functional projections. Concerning the second question (the position of obligatory movement of the subject), a possibility is that it simply is the position of (Nominative) Case assignment, although it remains to be understood why English and Italian vary in this respect from Norwegian. As to the third question (the apparent optionality of the movement of the subject around the higher adverbs), I have no interesting suggestion to offer. Perhaps, the subject raises to check a (distinct) additional feature in a higher Spec, or else the movement is motivated by (LF) concerns of relative scope between the subject and different adverbs (mediated by the necessity for the subject to enter a (secondary) predication relation with a particular adverb).'°

5.2 The positions of object DPs There is evidence that objects too can occupy several different positions. While in Danish and Norwegian “object shifted” pronominals must raise to a position to the left of all adverbs, “most varieties of Swedish permit Object Shift to intermediate positions as shown in [12]” (Holmberg and Platzack 1995, 153, n. 11). See also (13), from Holmberg (1993, 31, n. 11), and (14), from Holmberg and Platzack (1995, 161, n. 19): (12) De laser (den) troligen (den) girna (den) alla (?den). ‘They read (it) probably (it) with pleasure (it) all (it).’

(13) a. Numera g6r mej alltsa inte lingre Helge lika irriterad som forr. ‘Nowadays makes me thus no longer H. as irritated as before.’ b. Numera gor alltsa mej inte langre Helge lika irriterad som forr. c. Numera gor alltsa inte mej langre Helge lika irriterad som forr. d. Numera gor alltsa inte langre mej Helge lika irriterad som forr.

(14) Dom liste inte den alla. ‘They read not it all.’

As Verner Egerland informs me, a pronoun can indeed be found interspersed among the adverbs in the hierarchy of (8) (see (15)), but can be no lower than helt ‘completely’ (see (16)):!” (15) a. Jag gjorde drligt talat det lyckligtvis inte. ‘I did frankly it luckily not.’ b. Han sag lyckligtvis henne tydligen. ‘He saw luckily her evidently.’ c. Jag traffar sannolikt honom nu. ‘I meet probably him now.’ d. Han goér kanske det klokt nog. “He does perhaps it wisely.’ e. Han bjéd vanligtvis henne dterigen pa middag dagen efter. “He invited usually her again to dinner the day after.’ f. Jag triffar inte langre honom alltid (pa fredagar). ‘I meet no longer him always (on Friday).’ g. Han far ndstan det helt som han vill ha det. ‘He gets almost it completely as he likes it.’

116 ADVERBS AND FUNCTIONAL HEADS

(16) a. *?Han sag helt den. (cf. Han sag den helt.) ‘He saw completely it.’ b. *?Han sag tidigt den. (cf. Han sag den tidigt.) ‘He saw early it.’

Apparently, the same distribution is found with Norwegian full object DPs. Pronominals, as noted, “object shift” to a position preceding all adverbs, but full DPs can either precede or follow any adverb down to helt ‘completely’, though they precede (helt, preferably, and) godt ‘well’ and tidlig ‘early’. Some examples, provided by @ystein Nilsen, are given in (17) and (18):'8 (17) a. *Na liker Per heldigvis ikke lenger helt godt Jon. ‘Now likes P. fortunately not any.longer completely well J.’ b. *?Na liker Per heldigvis ikke lenger helt Jon godt. c. Na liker Per heldigvis ikke lenger Jon helt godt. d. Na liker Per heldigvis ikke Jon lenger helt godt. e. Na liker Per heldigvis Jon ikke lenger helt godt. f. Na liker Per Jon heldigvis ikke lenger helt godt.

(18) a. Deretter fortalte Per zrlig talt alltid denne historien helt til festdeltakerne. ‘Thereafter told P. honestly spoken always completely this story to the party participants.’ b. Deretter fortaite Per zerlig talt denne historien alltid helt til festdeltakerne. c. Deretter fortalte Per denne historien erlig talt alltid helt til festdeltakerne.

Although Swedish and Norwegian allow DPs, but not verbs, among the various adverb classes of (8), I will take such cases as (17) and (18) to indicate the presence of full DP-related projections among the Adverb-related ones.'”

5.3 Floating quantifiers A comparable situation is found with floating quantifiers (FQ’s), which I take, following Sportiche (1988), to be “stranded” in positions where DPs can stop (or transit)—that is, the specifier of DP-related projections.” I will start with French, which allows for subject, direct object, and indirect object ones, in the order FQs¢upjecty > FQiadirect) O(bject) > FQn(irect) O(bject).

That a FQ, must precede a FQ) is shown by (19):?! (19) a. Elles, leur, ont toutes; tous, parlé hier. they (FEM) to-them have all(FEM) all talked yesterday b. *Elles, leur, ont tous, toutes, parlé hier.

Examples (20a—b) show that a FQ, must precede a FQp0:”” (20) a. Les filles; les, ont toutes, tous, lu. the girlsCFEM) them have all(FEM) all read b. *Les filles; les, ont tous, toutes, lu.

As to the relative order between FQno and FQ,o, Kayne (1975, sec. 2.14) notes that in such examples as (21) (corresponding to his (276)) there is a unique interpretation, with the first quantifier tous necessarily relating to the dative clitic leur, and the second (toutes) relating to the accusative clitic les (as indicated by the variant with indices added):77

DP-RELATED FUNCTIONAL PROJECTIONS AND NEGATIVE PHRASES 117

(21) Je les leur ai tous toutes montrées. (?Je les, leur, ai tous,,+; toutes,4, montrées.) ‘I them to-them have all all shown.’

The grammaticality of (19a) and (20a) with crossing indices and (21) with nesting indices appears to suggest that there is no requirement on the way each pair of clitic and FQ relates to the other pairs (see Sportiche 1988, 435).*4 Comparable conclusions hold with respect to the order of FQ’s in Italian, except that Italian does not allow for a prepositionless FQ,9 (see ?Vi ho scritto *(a) tutti ‘I you have written all’).”° The position occupied in French by the prepositionless FQ;y appears to host 1n Italian the dative weak pronominal loro (see Cardinaletti 1991). As the contrast between (22a) and (22b) shows, lore appears to precede a FQOno, just as a FQ,o precedes FQpo in French:*° (22) a. (?)Li ha dati loro tutti GIANNI (non Mario). them has given to-them all G. (not M.) b. *Li ha dati tutti loro GIANNI (non Mario).

If we consider the order between loro and a FQg, at first sight we find that both relative orders are possible (with, in the case of (23b), a virtual pause before and after loro): (23) a. Hanno tutte loro dato tutto, (also: Hanno tutte dato loro tutto.) they (FEM) have all(FEM) to-them given everything b. Hanno loro tutte dato tutto. (also: Hanno loro dato tutte tutto.) they (FEM) have to-them all (FEM) given everything

Is it the FQ or is it loro that can occur in different positions (or both)? Certain observations seem to indicate that loro can indeed occupy two distinct positions. [f we insert an adverb like sempre, which we already saw occupies a fixed position, we observe that Joro may occur both to its left, and to its nght: (24) a. Han dato loro sempre tutto. they-have given to-them always everything b. Han dato sempre loro tutto. they-have given always to-them everything

Now, when loro is to the right of sempre, it has to follow the FQs.:7’ (25) a. ?7Han sempre tutte loro dato tutto. they-have always all to-them given everything b. *Han sempre loro tutte dato tutto. they-have always to-them all given everything

In the light of this, we interpret (23b) as having loro in the higher of the two positions (that to the left of sempre). As a matter of fact, the higher position which loro can occupy appears to be a sort of Scrambling position, which precedes the negative adverb mica (in fact, the whole sequence of adverbs in (31a)) and is typically open to PPs (but not DPs) containing a pronoun. See (26) and (27) (and n. 7 of this chapter): (26) a. Da aliora, non ho dato a lui mica pit sempre tutto. b. *Da allora, non ho dato mica a Jui pitt sempre tutto. c. *Da allora, non ho dato mica pit a lui sempre tutto.

118 ADVERBS AND FUNCTIONAL HEADS

d. *Da allora, non ho dato mica pit! sempre a Jui tutto. ‘Since then, I have not given to him no more always everything.’

(27) *Da allora, non ho invitato lui mica pit sempre. ‘Since then, I have not invited him no more always.’

Given the different positioning of the FQ, and the FQpo with respect to loro, we expect

the FQ, to precede the FQpo (as we saw it being the case in French). This seems essentially to be borne out by the facts. See (28)-(30), which are identical modulo the position of the past participle:”* (28) a. I bambini le avranno tutti tutte sistemate. the children(MASC) them(FEM) will have all(MASC) all(FEM) arranged b. *I bambini le avranno tutte tutti sistemate. the children(MASC) them(FEM) will have all(FEM) all(MASC) arranged (29) a. I bambini le avranno tutti sistemate tutte subito. the children(MASC) them(FEM) will have all(MASC) arranged all(FEM) immediately b. *?I bambini le avranno tutte sistemate tutti subito. the children(MASC) them(FEM) will have all(FEM) all(MASC) arranged immediately

(30) a. I bambini le avranno sistemate tutti tutte subito. the children(MASC) them(FEM) will have arranged ali(MASC) all(FEM) immediately b. *I bambini le avranno sistemate tutte tutti subito. the children(MASC) them(FEM) will have arranged all(FEM) ali(MASC) immediately

Having concluded that the order is FQ, > FQjo/loro > FQpo,”’ let us consider the location of such phrases within the adverbial sequence in (31): (31) a. mica> gia > pili > sempre > completamente > tutto > bene b. pas > déja > plus > toujours > complétement > tout > bien

Beginning with French, we note that apparently no FQ can intervene between bien and the past participle (see (32)-(34), again disregarding the “concessive,” nonmanner, usage of bien): (32) a. *Ils ont bien tous compris. ‘They have well all understood.’ b. Ils ont tous bien compris.

(33) a. *Je les ai bien tous reparés. ‘I them have well all repaired.’ b. Je les ai tous bien reparés.

(34) a. *Je leur ai bien tous répondu. ‘IT to-them have well all replied.’ b. Je leur ai tous bien répondu.

As noted in Kayne (1975, 156) a FQjo cannot intervene between tout and the past participle, but must precede tout: (35) a. *Je leur ai tout tous montré. ‘I showed them all everything.’ b. ?Je leur ai tous tout montré.

The same holds for a FQ FQig > FOpo > complétement > tout > bien

Similar facts hold im Italian. Neither FQ,’s, nor FQpo’s, nor /oro, can occur between bene and a complement of the verb:

(42) a. *Han rifatto bene tutti questo. ‘They have done again well all this.’ b. *Li ho spiegati bene tutti a Gianni. ‘I have explained well all to G.’ c. *Ha spiegato bene loro questo. ‘He explained well to-them this.’ Nor can they occur between completamente and bene:°* (43) a. *Han rifatto completamente tutti bene il compito. ‘They have done again all well the homework.’ b. *Li ho spiegati completamente tutti bene a Gianni. ‘T have explained them completely all well to G.’ c. *Ho spiegato completamente loro bene il compito. ‘T have explained completely to-them well the homework.’

Another FQ “space” is to the left of toujours, in between it and plus:°° (44) a. Ils n’ont plus tous toujours tout fait. ‘They not have no more all always everything done.’

120 ADVERBS AND FUNCTIONAL HEADS

b. Je ne les ai plus tous toujours invités. ‘I not them have no more all always invited.’ c. Je ne leur ai plus tous toujours tout dit. ‘I not to-them have no more all always everything said.’

Apparently, no FQ can occur in between déjd and plus: (45) *Ils n’ont déja tous plus rien recu. *They not have already all no more nothing received.’

although they can occur between pas and déja: (46) Ils n’ont pas tous déja téléphoné. ‘They not have not all already telephoned.’

The fact that FQ’s cannot be found preceding pas either (see (47)) suggests that, in their presence, there may be some general requirement on tous being in the (c-command) scope of negative pas and plus:*4 (47) *Ils n’ont tous pas toujours téléphoné. ‘They not have all not always telephoned.’

As in French, FQ’s (and loro) can appear between sempre and completamente, or before sempre: (48) a. Han rifatto (tutti) sempre (tutti) completamente il loro compito. ‘They have done again (all) always (all) completely their homework.’ b. Li ha rifatti (tutti) sempre (tutti) completamente bene. “He has done again (all) always (all) completely well.’ c. Ha (loro) sempre (loro) spiegato tutto. ‘He has (to-them) always (to-them) explained everything.’

Again, as in French, FQ’s (when under the scope of negation) cannot precede piu or mica (49a-b), but can precede, or follow, gia (50): (49) a. *Non hanno tutti mica pit accettato. ‘They not have all not any longer accepted.’ b. *Non li ho tutti mica pit invitati. ‘Not I them have all not any longer invited.’

(50) a. Hanno (tutti) gia (tutti) ricevuto tutto. ‘They have (all) already (all) received everything.’ b. Li ho (tutti) gia (tutti) sistemati bene. ‘I them have (all) already (all) accomodated well.’

Although it remains to be understood why not all positions among adverbs allow for the occurrence of FQs, these provide evidence for the existence of at least several DP-related projections among the Adverb-related projections.

5.4 The positions of Neg(ative) P(hrase)s In this section, I review some evidence suggesting that negation can also occur in several distinct positions in the clause. These positions appear sometimes realized simultaneously (in certain cases contributing a single instance of negation, in others contributing multiple negations, which cancel each other out). In a detailed study of negation in Romance, Zanuttini (1997) strongly argues for the existence of (at least) four distinct positions where NegPs can be generated

DP-RELATED FUNCTIONAL PROJECTIONS AND NEGATIVE PHRASES 121

within the clause: one below C°, hosting a negative head (Italian non, Paduan no,

etc.), and three NegPs interspersed among “lower” AdvPs, hosting a negative XP in specifier position. Of the latter three, one is immediately above the projection hosting “already” (where we find Italian mica, Catalan pas, Piedmontese and Valdotain pa, Milanese minga and Pavese mia: all presuppositional negations; see chapter 1 in this volume and Zanuttini, forthcoming, chap. 3); another is in

between the projection hosting “already” and that hosting “no longer” (where we find Piedmontese nen, and Valdotain non-presuppositional pa); and the third below the projection hosting “always” (where we find Milanese and Pavese no, both non-presuppositional).*°

As shown in (51), adapted from (99) of Zanuttini (1997, 88), a number of Romance varities appear to utilize more than one such position (sometimes simultaneously):*° (S1)

NegPl NegP2 already NegP3 nolonger always NegP4

Italian non... mica gia 0 pil sempre 0

Pavese 0... mia giamo6 =O pil sempar no Milan. 0... minga gemod 0) pi semper no

Piedm. 0... pa gia nen pi nen sempre 0 Valdo. OQ... pa dza pa pa mai toujou 0 French 0... pas déja 0 plus toujours guére?” One of the questions that arise is whether a NegP should be assumed to be structurally present even when it contains no overt material. Provisionally, I will assume that it is not, and that the different positioning of NegP, including the higher positioning to be reviewed directly, is principally a function of the different scope of negation relative to other elements.** This is already suggested by Zanuttini (1997) for NegP2, NegP3, and NegP4. For example, she notes that Piedmontese nen (in NegP3) cannot take scope over “already,” which it fails to c-command. Pa or nen in NegP2 (a possibility only marginally admitted for the latter), instead can, as the Spec of NegP2 c-commands “already.” The variable positioning of negation in the higher portion of the clause seems to depend (at least in part) on similar scope concerns. This requires, however, that we take a closer look at the nature of the higher instances of negation. Zanuttini (1997, chap. 2) argues that preverbal negation in Romance is essentially of two kinds (a head, in both cases). The one that cannot negate the sentence by itself (but co-occurs with a postverbal negation) is cliticized to the (finite) verb, and can in certain cases surface after subject and complement clitics; the other, which can negate the sentence by itself (and is thus the real negation) heads a NegP (NegP1), immediately below C°. Here I will depart from the latter conclusion and assume that even the “real” negation can cliticize to the finite verb (or to the clitic which is cliticized to the finite verb).*? The main grounds for this conclusion are (1) the possibility, even for this negation, to be “carried along” by the verb to a C° position; and (2) the evidence that this negation originates not in a single NegP (NegP1), but in several different NegP positions, interspersed among the higher adverb-related projections (being subse-

122 ADVERBS AND FUNCTIONAL HEADS

quently carried along by the verb to different positions in the IP “space”). Consider the two issues in turn. Both in Italian and in Paduan, which utilize a preverbal negation which can negate the sentence by itself, the negation can apparently raise to C° together with the verb in the Aux-to-COMP constructions studied in Rizzi (1981, 1982). See (52):*° (52) a. Non essendo Gianni riuscito a ripararlo, dovemmo trovare un’altra soluzione. ‘Not having G. managed to repair it, we had to find another solution.’ b. (Va casa...) No fusse-lo gnancora ‘riva’, spete-lo fora.*! ‘(Go home .. .) Should he have not arrived yet, wait for him outside.’

Consider now the evidence for assuming that preverbal negation may originate in several different positions. Earlier | mentioned Zanuttini’s observation that postverbal negation in Romance cannot take scope over AdvPs preceding it (i.e., outside its c-domain). When we turn to preverbal negation a more complex picture emerges. While it remains true that (unmoved) AdvPs preceding it cannot fall under the scope of preverbal negation (53), in the case of AdvPs following it, we find two distinct cases. Some can be interpreted as either falling or not falling under its scope (54), while others never fall under it (55): (53) a. Francamente non ho altro da aggiungere. ‘Frankly I haven’t anything else to add.’ b. Gianni fortunatamente non é riuscito a corrompervi. ‘G. luckily hasn’t managed to corrupt you.’ c. Gianni evidentemente non ha saputo del nostro arrivo. ‘G. evidently hasn’t learned of our arrival.’ d. Gianni probabilmente non ¢ in grado di atutarci. ‘G. probably isn’t able to help us.’ e. Lei allora non poteva/potra aiutarci. ‘She then couldn’t/won’t help us.’ f. Forse non ha fatto una gaffe. “Perhaps he/she hasn’t made a blunder.’ g. Questa clausola necessariamente non avra conseguenze.** ‘This condition necessarily will not have consequences.’ h. Gianni deliberatamente non ha lasciato cadere la sua candidatura. ‘G. deliberately has not dropped his candidature.’ i. In questo paese, i negozianti obbligatoriamente non rilasciano ricevute. ‘In this country, shopkeepers obligatorily do not issue receipts.’ 1. Gianni astutamente non si é nascosto. ‘G. smartly did not hide.’ m. Gianni di solito non era disposto a compromessi. ‘G. usually was not disposed towards compromises.’ n. Gianni di nuovo non é arrivato puntuale. ‘G. again did not arrive punctually.’ o. Gianni spesso non ci lasciava il suo recapito. ‘G. often did not leave us his address.’ p. Gianni rapidamente non accetto la loro offerta. ‘G. rapidly didn’t accept their offer.’ q. Gianni gia non riceveva notizte da casa. ‘G. already did not receive news from home.’

DP-RELATED FUNCTIONAL PROJECTIONS AND NEGATIVE PHRASES 123

r. ?Gianni sempre non accoglieva le loro offerte di tregua. ‘G. always wouldn’t accept their offers for a truce.’ s. Gianni quasi non si era ricordato di mangiare. ‘G. almost didn’t remember to eat.’ ?Gianni parzialmente non ricordo le istruzioni. ‘G. partially didn’t remember the instructions.’

(54) a. Lei non poteva/potra allora aiutarci. ‘She couldn’t/won’t then help us.’ b. Non ha forse fatto una gaffe.*? ‘He/she hasn’t perhaps made a blunder.’ c. Questa clausola non avra necessariamente conseguenze. ‘This condition will not necessarily have consequences.’ d. Gianni non ha deliberatamente lasciato cadere la sua candidatura. ‘G. has not deliberately dropped his candidature.’ e. In questo paese, i negozianti non rilasciano obbligatoriamente ricevute. ‘In this country, shopkeepers do not obligatorily issue receipts.’ f. Gianni non si é astutamente nascosto. ‘G. didn’t smartly hide.’ g. Gianni non era di solito disposto a compromessi. ‘G. wasn’t usually disposed towards compromises.’ h. Gianni non é di nuovo arrivato puntuale. ‘G,. didn’t again arrive punctually.’ i. Gianni non ci lasciava spesso il suo recapito. ‘G. didn’t often leave us his address.’ 1. Gianni non accettO rapidamente la loro offerta. "G. didn’t rapidly accept their offer.’ m. Gianni non riceveva gia notizie da casa. ‘G. did not already receive news from home.’ n. ?Gianni non accoglieva sempre le loro offerte di tregua. °G. wouldn’t always accept their offers for a truce.’ o. Gianni non si era quasi ricordato di mangiare. *G. didn’t almost remember to eat.’ p. Gianni non ricord6 parzialmente le istruzioni. ‘G. didn’t partially remember the instructions.’

In (54), even if, depending on context, one or the other interpretation may be favored, both the interpretation in which the adverb is under the scope of negation, and the one in which it is not, are possible.¥ In (55), on the other hand, the AdvP is necessarily outside the scope of negation:® (55) a. Non ho francamente altro da aggiungere. ‘I haven’t frankly anything else to add.’ b. Gianni non é fortunatamente riuscito a corrompervi. “G. hasn’t luckily managed to corrupt you.’ c. Gianni non ha evidentemente saputo del nostro arrivo. “G. hasn’t evidently learned of our arrival.’ d. Gianni non é probabilmente in grado di aiutarci. ‘G. isn’t probably able to help us.’

The adverb classes in (55) interestingly correspond to a contiguous portion of the highest projections of the clause (from Modepistemic CO MOOdgpeech act» 1 the analysis of

124 ADVERBS AND FUNCTIONAL HEADS

the previous chapters). This fact, combined with the pattern found in (53) and (54), suggests to me the following conclusions about preverbal negation (in Italian):

1. What is crucial for determining the scope of sentence negation is not its “surface” position (the one at “Spell-out’), but its “base” position. If the former counted, we would expect negation to necessarily take scope over the AdvPs in (54) and (55), contrary to fact.* 2. Apparently, a NegP can be “base generated” on top of any of the Adverbrelated projections below Mod, icemic, With non, in Italian, originating in the position of the specifier, from where it cliticizes on to the verb raised to the head immediately above the NegP in question:*’ (56) ...lyplyp ly V-Y¥ ] [negep [ [non] |] [ negp Ll xp...

From these assumptions, the facts observed here concerning the interaction of adverbs and negation follow. When non (cliticized to the verb) 1s to the right of a certain adverb, it can only have originated in a NegP lower than the adverb (whence its not taking the adverb in its scope (53)).48 When non (cliticized to the verb) is to the left of a certain adverb (of the “space” below Mod,pistemc), We do not Know whether it originated in a NegP to the left of the adverb, or in one to its right (then, ending up to its left, on the verb’s shoulders, so to speak). This means that a sentence such as (54d), repeated in (57), may be derived either as shown in (58a) or in (58b) (whence the observed ambiguity of such cases): (57) Gianni non ha deliberatamente lasciato cadere la sua candidatura.

(58) a. ...[xplxpnon ha [negp[ Lt} ] [ negp t [ ypdeliberatamente [ yp. . [. .

b. ...[xp[xpnon ha [ ypdeliberatamente [ yp non ha [ negp | Lt] | Lnegp t L -

In the presence of two adverbs to the right of non + V it can be the case that only the rightmost is under the scope of negation, or both of them, or neither of them, but not just the leftmost. See (59): (59) Gianni non era deliberatamente di solito polemico con i suoi superiori.*” ‘G. wasn’t deliberately usually polemical with his bosses.’

This suggests that non can originate in a NegP in between the two adverbs, or to the right of both, or to the left of both (and similarly if more adverbs are added). And this entails that it must be possible to “base generate” a NegP on top of each adverb-related projection (within a certain “space’’). The fact that the highest adverbs of (55) cannot be interpreted under the scope of negation even when non+V precedes them appears to indicate that no NegP is available above a certain projection (which I have identified with ModPyisemic: 1 Whose Spec probabilmente ‘probably’ is generated). This accords well with the observation that epistemic modals are never under the scope of negation.”

DP-RELATED FUNCTIONAL PROJECTIONS AND NEGATIVE PHRASES 125

Languages where negation does not cliticize to (thus ‘riding on’) other elements moving up the clause (where, for example, it is a suffix, or a particle or a specifier)

appear to differ as to the canonical position they reserve to it among the Adverbrelated functional projections.°! We have already seen the variation uncovered by Zanuttini (1997) in the postverbal negations of Romance. Many languages have negation higher than T(Past) (see, for example, (60a—c)); others between T(Past) and

T(Future) (see (61) and (62)); others below T(Past) and T(Future) (see (63)); and other possibilities exist:°¢ (60) a. Pooy-ITTUND-aayir-unn-illa. (Malayalam; see Appendix 2) go-ANT-be.PAST-unn-NEG ‘(He) had not gone.’ b. A-ka-d-dzie-ak-A ... ngamba west. (Babele; Leitch 1994, 197) 3sg-NEG-PAST-eat-ASP-Final Vowel elephant all ‘He didn’t eat the whole elephant.’ c. xan ir-son—gui. (Mongolian; Svantesson 1991, 192) king come-PAST-NEG “The king did not come.’

(61) a. Pd na’ ké kwé mbé wa. (Bangwa; Nguendjio 1992, 95) Ipl PAST NEG eat-ANT meat NEG “We had not eaten meat.’ b. P35 k6 i nkwé mbé ws. Ipl NEG FUT eat-IMPERF meat NEG “We will not eat meat.’

(62) a. Ai bin no de du mai wok dat taim. (Nigerian Pidgin; Ofuani 1982, 236) I PAST NEG PROG do my work at that time b. Ai no go foget. (Nigerian Pidgin; Ofuani 1981, 315) I NEG FUT forget

(63) a. Calis-ma-yacak-ti. (Turkish; van Schaaik 1994, 39) work-NEG-FUT-PAST ‘(S)he wouldn’t work.’ b. (Pe ne-sitom re save) yoko naga ve vimi re poli. (Lewo; Early 1994, 74) (Neg! Isg-think Neg2 that) FUT he Negi come Neg2 Neg3 “(I didn’t think that) he will not come.’

It remains to be seen whether there can indeed be a NegP over every single adverbrelated functional projection below Mod, pistemice Some gaps exist. For example, neither “retrospective” adverbs (appena ‘just’) nor “durative” adverbs (brevemente ‘briefly’) can apparently be outside the scope of negation even though the lower “proximative” (quasi ‘almost’) and “completive” (parzialmente ‘partially’) adverbs can. Compare (64) with (65): (64) a. Non si é appena presentato. (See *Appena non si é presentato.) “He hasn’t just shown up.’ b. Non ha brevemente parlato. (See *Brevemente non ha parlato.)*? ‘He hasn’t briefly spoken.’

(65) a. Non si é quasi ricordato di venire. (See Quasi non si é ricordato di venire.) ‘He hasn’t almost remembered to come.’

126 ADVERBS AND FUNCTIONAL HEADS

b. Non si ricord6 parzialmente di quello che aveva fatto. “He didn’t partially remember what he had done.’ (See Parzialmente non si ricordd di quello che aveva fatto.)

Examples (64a—b) could, however, be “accidental gaps” due to the oddness of modi-

fying a negative proposition with such adverbs. The possibility of “base generating” a NegP in several positions interspersed among the higher (adverb-related) functional projections is also supported by the simultaneous realization of several negations which cancel each other out. See (66), (67), and (68), where arguably the 2’t and nae of (67) and (68) are heads and the not and no of (67) and (68) are specifiers (see Kayne 1989b, sec. V):

(66) Non é non andato distrutto. ‘It hasn’t not gone destroyed.’ (67) He couldn’t (possibly) not have accepted. (Richard Kayne, personal communication)

(68) He couldnae have no been no working. (Hawick Scots; Brown 1992, 84) ‘It is impossible that he has not been out of work.’

The fact that sentences containing more than three negations (canceling each other out) are in general unacceptable may be due to processing difficulties rather than to the impossibility of generating more than three NegPs. In conclusion, the evidence points to the possibility of generating a NegP on top of every adverb-related functional projection, even simultaneously, up to a certain height (which is likely determined by semantic reasons). On the nature of the Negative Phrase, it is occasionally proposed that negation and (emphatic) affirmation are the two values of one and the same phrase: a Polarity Phrase. This, however, is dubious, quite apart from the fact that both appear to be marked values. The reason is that they can (marginally) co-occur in a fixed relative order (in Italian), with emphatic affirmation (si?) preceding negation (mica): (69) a. (?) Gianni non ci ha si mica detto tutto (ma ce Jo ha lasciato capire). ‘G. not us has yes not told everything (but he has let us understand it).’ b. *Gianni non ci ha mica si detto tutto (ma ce lo ha lasciato capire).

If anything, this suggests the existence of two separate projections. Emphatic affirmation also seems to occur in various positions (see Poletto, 1998).

6

Some Implications and Residual Questions

In this chapter, I briefly discuss some implications that derive from the preceding discussion, and point to certain areas in need of further investigation. The conjecture that AdvPs are specifiers of distinct functional heads, entering with them a transparent semantic relation, raises a number of issues. One 1s the question of parameterization. How much variation should UG allow in the number and type of functional projections available to different languages? The strongest position would be that UG allows no variation at all. The evidence discussed above seems to suggest that this position may actually be tenable. Languages do not seem to differ as to whether they have aspectual projections higher or lower than mood projections, epistemic modality higher or lower than root modality, etc. The order appears cross-linguistically invariant. The limited cases of apparent variation (§6.4) all seem to involve agreement and negation in relation to other functional heads. But we have seen that precisely AgrPs (rather, DP-related projections) and NegPs are generable in many different positions among the adverb-related functional projections. It is thus tempting to interpret such variation as stemming from a pure “spell-out” option: whether a language lexicalizes a higher or lower Agr or Neg. A related question is whether we should take the entire array of functional projections to be present in every

sentence. I will suggest that this is the least costly assumption, once we recognize that each head comes with a marked and a default value (see $6.1). This conclusion, if correct, opens up a new view of clausal structure—one that is further removed from what we see, but no less interesting, for that. If a universal hierarchy of functional projections exists, one may ask whether it is primitive, or itself determined by more abstract principles, plausibly semantic prin127

128 ADVERBS AND FUNCTIONAL HEADS

ciples. Although the question perhaps cannot be meaningfully answered at this point,

some slight evidence exists that the hierarchy may be a primitive property of the computational component, only indirectly related to semantic considerations (see §6.3).

If these conclusions are correct, many more questions arise which will need further investigation. Among these, the nature and parameterization of DP- and V-movements along the hierarchy of adverb-related functional projections.

6.1 Default and marked values: simple and complex sentences Let us consider in more detail the possibility that each adverb-related functional projection comes with two values: an “unmarked” (or “default”) value, and a “marked” one. The notion of markedness, used prominently in the structuralist and early generative traditions to mark the members of an opposition, is usually characterized by a number of different criteria (which ideally should converge; Moravcsik and Wirth 1986).! “Marked” is the member with the more restricted application (bitch—only female—versus dog), less frequent, conceptually more complex, expressed by overt morphology; “unmarked” the member with wider application (dog—male, or unspecified for male/female), more frequent, conceptually basic, often expressed with zero morphology (see Greenberg 1966, 25ff, for discussion, and additional criteria). Here, I will not attempt a more precise definition, but simply follow the tradition in its characterization of various functional notions on the basis of these criteria. Two observations of Jakobson’s will be particularly valuable in deciding which of two values is marked when the tradition is silent on the question. The first is his characterization of “unmarked” as ambiguous (or wider) in application: “The general meaning of a marked category states the presence of a certain (whether positive or negative) property A; the general meaning of the corresponding unmarked category states nothing about the presence of A, and is used chiefly, but not exclusively, to indicate the absence of A” (Jakobson 1957/1971, 136). The second 1s his observation that zero morphology typically occurs with the unmarked members of categories (Jakobson 1939/ 1971; also see Comrie 1976a, 114; Bybee 1985, 52ff). In the area of illocutionary force (or speech act) distinctions, for example, it is traditionally assumed that “declarative” (or “indicative’’) is the unmarked form, all other choices (interrogative, imperative, hortative, etc.) being marked (Bybee 1985, 147; Lyons 1968, 307). Descending the hierarchy, I consider the “negative” value of the evaluative mood head (unfortunately) to be the marked value and the unspecified, or the “positive,” value to be the default value.* Concerning the evidential mood head, I consider direct (visual) evidence to be the default value and all other kinds of evidence to be marked. This decision appears supported by the fact that many languages making evidential distinctions have no overt morphological marker corresponding to direct (visual) evidence (as opposed to other kinds of evidence). This is true of the Germanic and Romance languages mentioned in §4.6, as well as for Makah (Jacobsen 1986, 9), Wintu, Maidu, Ute, and other languages (Willet 1988, 64ff). For the epistemic modal head, I consider the case in which the speaker is committed to the truth of the

SOME IMPLICATIONS AND RESIDUAL QUESTIONS 129

proposition to be the “default” value. This value is implicit in every statement, and must be otherwise explicitly denied (with probably, presumably, etc.), when the speaker does not want to commit himself or herself. Concerning the various tense heads (T(Past), T(Future) and T(Anterior)), fol-

lowing Vikner (1985, 94), I consider the case where the time points coincide (R,,S;R>.,R);E,R,) to represent the default value, and that where they do not (R,_S;R,_R,;E_R.) to represent the marked value. Usually it 1s the noncoincidence, not the coincidence, of the time points that is morphologically marked (e.g., -ed and will versus 0, in English). Again, following the tradition, I consider “realis” to be the default value of the “realis/irrealis” mood head, which is located below T(Future). Consider now alethic and root modals. Here, I consider the marked values to be [ — necessary], [ — possible], [ — volition], [ — obligation], [ — ability/permission], and the default values to be — [ — necessary], etc.> Concerning the aspectual heads, I con-

sider the presence of the relevant feature (which is typically associated with overt morphology) to represent the marked value of the head and its absence to represent the default value. The marked versus default values of the various aspectual heads seen here are thus: (1) [+ habitual] versus — [{ + habitual], [ + repetitive] versus — [ + repetitive], | + frequentative] versus — [ + frequentative], [ + celerative] versus — | + celerative], | + terminated] versus — [ + terminated], [ + continuative] versus — [ + continuative], [ + perfect] versus — [ + perfect] (or imperfect), [ + retrospective] versus — [ + retrospective], [ + proximative] versus — [ + proximative], [ + durative] versus — [ + durative], [| + progressive (or temporally specific)] versus ~ [ + progressive] (or temporally generic), [ + prospective] versus — [ + prospective], [ + completive] versus — [ + completive]. Those concerning Voice are [ — active] ( = passive) versus — [ — active] ( = active).

This oversimplifies things in certain ways. Yet I will keep to it for convenience here. As is noted in the literature (Comrie 1976a, §6.5; Bybee 1985, 147), the relevant notion of markedness is not absolute but relative (or “Jocal”). So, for example, while in the presence of — Past the unmarked value is — [ + terminated], in the presence of + Past the unmarked value appears to be [ + terminated]. Also, while — [ + completive] is apparently the unmarked value with activities and states, it seems that [ + completive] is the unmarked value for telic predicates (so that, as noted in chapter 4, a completive interpretation is invited in such cases as John has eaten the sandwich, unless explicitly denied: John has partially eaten the sandwich).4 Such interrelations with “lexical aspect,” or in the different values of the functional heads among themselves, remain outside the scope of this work. In any event, they do not call into question the distinction between marked and default values of a head. They only show that in the presence of one or another lexical class of predicates, or of a particular value in another functional head, what counts as the default or marked value may shift (for reasons that remain to be elucidated). Certainly, the whole issue of markedness would deserve a more careful examination, but this discussion, summarized in Table 6-1, will suffice for present purposes, namely for the “logic” of the question.

Table 6.1 Default and marked values of the different functional heads. (Recall the possible different locations of the root modals with regard to other heads, possibly much lower than indicated here; cf. chapter 4 and later in this chapter.)

Functional head Default Marked Mood, peech act declarative — declarative

Mood, yatuative — [| — fortunate] — fortunate Mood evidential direct evidence — direct evidence

Modepistemic committment — committment

T(Past) Ris R,_S T(Future) R,,R, R,_R>

Mood, reatis realis irrealis

Mod,jeth necess — [| — necessary] — necessary

Modatetn possib — [ — possible] — possible Modyolition — [ — volition] — volition Mod bigation ~ [ — obligation] — obligation

Mod binty/permss — [ — ability/permission] — ability/permission

ASPhabitual — [ + habitual] + habitual

ASP yepetitive(!) —[ + repetitive] + repetitive ASP frequentative() — [| + frequentative] + frequentative ASD celerative() — { + celerative] + celerative

T(Anterior) E,R> E_R, ASD termmative — { + terminative] + terminative

ASP continuative — [ + continuative] + continuative

ASD perfect imperfect perfect

ASP retrospective — [ + retrospective] + retrospective ASP proximative — [ + proximative } + proximative

ASP gurative — [ + durative] + durative

ASP progressive generic progressive ASP prospective — [ + prospective] + prospective ASP completiveSg — [ + completive] + completive ASP completiveP| — [ + completive] + completive

Voice active passive

ASD celerativeD) — [+ celerative] + celerative

ASP repetitive) —[ + repetitive] + repetitive

ASP frequentativ(lt) — | + frequentative] + frequentative

ASP completive(I) — [+ completive] + completive

SOME IMPLICATIONS AND RESIDUAL QUESTIONS 131

The fact that functional heads come with two values (granting now the plausibility of this idea) has consequences for a number of more general questions. One,

which I address in the rest of this section, is the relation between simple and compound tense sentences. (Other implications will be discussed in the following sections.) A fairly widespread assumption is that compound tense sentences are substantially richer in functional structure than simple tense sentences, to make room for the extra grammatical elements (auxiliaries, particles, etc.) present in one but not the other. If functional heads necessarily come with a default and a marked value, this is an illusion. A simple sentence such as (2a) would have exactly the same functional structure as the apparently much richer (2b):

(2) a. Prices rise. b. Prices must not have been being raised.

From this perspective, the only difference between (2a) and (2b) lies in the presence of more morphology in the latter, due to the association (in English) of particular morphemes with the marked values of some of the functional heads. Example (2b) has marked values for Voice, AsPprogressivey ASPperfect» Neg, MOdepistemics While (2a) has

the corresponding default values). The conclusion that (2a) may not be fundamentally different in functional structure from (2b) is rendered plausible by the observation that concerning it we know that it is active rather than passive, generic rather than progressive, positive rather than negative, declarative rather than interrogative, etc.: implicit, not inexistent, information. Another possibility would be to say that the functional make up of (2a) is “read off” from an impoverished structure as a series of default values (see Giorgi and Pianesi, 1997, chap. |). I return to this alternative in the next section, concluding that it compares unfavorably with the idea that (2a) has a full functional structure, like that of (2b). Similar remarks will hold for the relation between (2a) and (3), which, in this perspective, does not involve more (adjunction, or specifier) structure:° (3) Fortunately prices probably already no longer rise so rapidly.

This should not be taken as saying that the lexical structure of (2b) is necessarily the same as that in (2a). At first sight 1t might be tempting to assume that the various aspectual and temporal morphemes (say perfect -en, and future will, in English) are directly generated under the corresponding functional heads, with auxiliaries generated in higher heads still, when required to bear morphology that would otherwise remain stranded (or to check higher features).° Whether this is correct, wholly or in part, remains to be ascertained. An apparent problem is that in certain cases such morphology apparently occurs in positions lower than the position occupied by the corresponding functional heads. As already noted in Baker (1971, 167ff and n.1), under normal conditions, auxiliaries and modals are unstressed and appear to the left of all adverbs. Some, however, can follow (most) adverbs when they bear nonlow stress (which happens in two cases: “The first is that in which the finite auxiliary is emphasized, the second

132 ADVERBS AND FUNCTIONAL HEADS

that in which the constituent following the auxiliary has been deleted” (Baker 1971, 71) {or removed]). See (4a—b) (= Baker’s 1971, (8b) and (8c), and (5a—b) (= Baker’s 1989, chap. 11, (57)): (4) a. John admires Susan now, and he always HAS admired her. b. *John admires Susan now, and he HAS always admired her.’

(5) a. Bill never will . b. *Bill will never , Example (Sa), with the future modal will following a relatively low adverb, suggests that the modal does not occupy the (much higher) T(future) head.

In general, it appears that along the hierarchy of adverbs seen here certain modals and auxiliaries can remain as low as the position to the immediate right of almost (or higher), but not any lower (see (6)—(9)). Perhaps this indicates that they are generated there Gf will can also occur in the Prospective Aspect head, and root can/could are in a head below it), but more work is needed before any conclusion can be reached.® (6) He said he would destroy it a. *?.., and he completely will/has. b. *?... and he will/has completely. (7) He said he would do his homework somehow by tonight

a. *...and he well will/has. b. *...and he will/has well. (8) He said he would wake up by himself

a. *...and he early will/has. b. *...and he will/has early. (9) He said he would finish his homework by tonight?’

a. *...and he fast has/will. b. *...and he has/will fast.

6.2. The hierarchy of functional projections and minimalist ideas The idea that all languages have the entire array of functional projections available (say on the order of 40), and that, perhaps, even simple sentences necessarily instantiate it, may seem to be in stark contrast with the minimalist spirit of Chomsky (1995). The contrast, however, is more apparent than real. As Chomsky says, “postulation of a functional category has to be justified, either by output conditions (phonetic and semantic interpretation) or by theory-internal arguments” (p. 240). My argument all along has been that each projection has a specific semantic interpretation. The crucial point, then, is whether all languages have the same full array of interpretations. Although we cannot be certain, as usual, the available evidence indicates that they do. Language after language, the same (restricted) functional notions appear to be expressible either via head morphology or via adverbs, while other, conceivable, functional notions are not.!°

SOME IMPLICATIONS AND RESIDUAL QUESTIONS 133

Acknowledging that all languages have at their disposal the same full array of functional notions,!! the next question is whether they utilize it all the time. Here the answer must be more tentative. I have already suggested that if heads necessarily come with a marked and a default value, then it is conceivable that all sentences utilize the entire functional structure, with the required combination of marked and default choices. But other possibilities exist. For instance, Giorgi and Pianesi (1996; 1997, chap. 1, §1.4) propose, in a “minimalist” spirit, that even though a language may have access to the maximal number of functional projections made available by UG, it will each time utilize only those projections that are needed to host specific lexical or morphological material present in the numeration (in Chomsky’s 1995 sense). !4

So, for example, where there is neither head morphology nor adverbs corresponding to a particular functional projection, that projection will not be projected, and its features (required for interpretation at the LF interface) will be syncretically projected with other features that are structurally represented for morphological or lexical reasons. What is required in such a system is a general convention that interprets the absence of a certain projection as corresponding to the default value of the relevant head. Though attractive, I think that such idea is more costly than the idea that functional notions are always all structurally represented.

Consider the case in which a projection must be projected (in their system) because there is an adverb in the numeration corresponding to a certain functional feature. The appropriate structure is required even if the adverb corresponds to the default value of the head (e.g., John has completely eaten the sandwich).'? This, however, means that the default value is once “read off” the actual structure (when the adverb is present), and once provided, via a general convention, when no adverb (hence no structure) is present. If the entire functional structure is always present, one can “read off” the default value (or the marked value, for that matter) once and for all from the structure (whether or not an adverb is present). In particular, no additional convention is needed for the default case. Consider now a more radical “minimalist” alternative; one that doesn’t necessarily project adverbs in the specifier position of a corresponding functional projection, but adjoins them to other projections. Although such system cannot order the adverbs by exploiting the principle (whatever that is) which determines the relative scope of functional projections, it can have an independent scope principle for adverbs (comparable in cost to that for functional projections). From this perspective, the two alternatives would seem to be equivalent (the second being more parsimonious). Certain observations made in chapter 2, however, point to the advantage of deriving the effects of the scope principle for adverbs from the different specifier positions the adverbs occupy in a structure representing the full array of functional projections. We saw that the same verbal form (say, an active past participle) is ordered differently with regard to the same adverbs in different Romance varieties, and that distinct verbal forms (say, an active past participle and an infinitive) are ordered differently with regard to adverbs within the same variety. For example, we saw that in French an active past participle is to the right of ‘well’; in Sardinian to the left of ‘well’ and to the right of ‘everything’; in Italian to the left of ‘everything’ (and ‘well’)

134 ADVERBS AND FUNCTIONAL HEADS

and to the right of ‘completely’; in Bellunese to the left of ‘completely’ and to the right of ‘always’.'* Similarly, we saw, in the case of French, that while an active past participle cannot be to the left of ‘well’, an infinitive can (see Pollock 1989). Similar facts are found throughout Romance (see chapter 2 and Zanuttini, 1997, chap. 3). In a system where adverbs are (left-)adjoined to the (single) projection hosting an active past participle or to an XP to its right, one must add a specific instruction for each adverb (whether it can be to the left or to the right of the past participle). The particular implications that were observed in chapter 2 (if an active past participle—more generally, any verbal form—can be to the left of adverb, it can be to the left of adverb;, adverb,, etc.) cannot be stated simply and revealingly. The scope principle governing the relative scope of two, or more, adverbs has nothing to say in this connection, because here it is the relative position of the verb with a single adverb that matters. Even if we were to say that when a certain adverb can be (left-)adjoined to the right of the projection hosting the past participle, then all adverbs with narrower scope must do likewise, we would still not have a principled reason why that is the case. And we wouldn’t have a reason why an adverb with narrowest scope cannot (left-)adjoin to the left of the projection hosting the past participle. After all, only the position of one adverb is at stake. The representation of the relative scope of adverbs in structural terms, in a con-

figuration also representing the position of the active past participle, can express the aforementioned implications in a very simple and restrictive manner. If we assume that adverbs are “base generated” in different specifier positions according to a certain principle of relative scope (which is the same for the corresponding heads), and we assume that the active past participle moves (obligatorily) to different head positions depending on the specific Romance variety, nothing else needs to be said. It follows that when the past participle precedes a certain adverb it will necessarily precede all adverbs lower in the (scope) hierarchy, which are also lower in the structure. The same argument can be replicated for each verbal form. Unless we opt for a structural representation of scope, where the adverbs are specifiers of projections to whose heads the verbal form can or cannot raise, we will be unable to express the simple generalizations linking together the different adverb classes and the different verbal forms. Why should an adverb that cannot (Jeft-)adjoin to the right of a past participle (left-)adjoin to the right of a finite verb, or an infinitival? In the fully structural hypothesis, if we grant that infinitivals (usually) raise higher than active past participles, that finite verbs raise higher than infinitivals (and that auxiliaries raise higher than lexical verbs of the same form) nothing else needs to be said.

6.3 Semantics and the hierarchy of functional projections If clausal functional notions such as Mood, Tense, and Aspect indeed prove to be arranged in a rigidly fixed hierarchy universally, a legitimate question is whether such

a hierarchy is primitive or rather determined by higher order (plausibly semantic) conditions. [f such functional elements were conceived of as logical operators on the

predicate-argument structure of the clause, their entering a certain relative scope

SOME IMPLICATIONS AND RESIDUAL QUESTIONS 135

among each other would be entirely natural. In this view, their order would simply reflect the intrinsic logical ordering of the operators. Certain considerations suggest that while the relative order of some such notions may indeed reflect intrinsic logical relations among them, the hierarchy is only indirectly related to such semantic, or logical, properties. For example, certain possibilities which in terms of logical relative scope would be expected are not found, or are downright impossible, thus suggesting that the hierarchy is a construct of the computational system of language, not completely reducible to other components. Although such conclusion may be due to my lack of knowledge or imagination, I will keep to it up to disconfirmation. The fact seen in chapters 3 and 4 that epistemic modality is higher than (takes scope over) Past, or Future, Tense (see also Bybee 1985, 119ff) appears to reflect the intrinsic relative scope of the two types of operators. Epistemic modality, as noted, expresses

the degree of the speaker’s commitment to the truth of the proposition expressed by the sentence. But the truth of the proposition can only be evaluated if the proposition is located in a precise moment of time. Consequently, epistemic modality presupposes an operand which is already tensed (whence its scope external to Tense). Similarly, as remarked in Foley and Van Valin (1984, 209ff), aspect is more closely related to the predicate (it expresses different ways of viewing the event expressed by the predicate) than tense (which locates the time of the event-——whatever its aspectual make up—with respect to the speech time). As a consequence of that, it is natural that tense operates on something on which aspect has already operated; in other words, that it is external to (higher than) aspect. Other such cases of “intrinsic” ordering exist. But, not all cases of relative order among functional elements seem so explicable. Consider the relation between Evidentiality and Epistemic Modality. In the previous chapters, it was noted that evidential heads (affixes or free morphemes) and specifiers (AdvPs) are higher than epistemic ones. Their relative order cannot be subverted. See, for example, (10a—b) and (11la—b):

(10) a. Allegedly John will probably give up. b. *Probably John will allegedly give up. (11) a. (?) Evidentemente Gianni ha probabilmente lasciato Il’ albergo. ‘Evidently G. has probably left the hotel.’ b. *Probabilmente Gianni ha evidentemente lasciato 1’ albergo. ‘Probably G. has evidently left the hotel.’

It is not clear whether this fact derives from any intrinsic logical ban on epistemic modality taking evidentiality in its scope. See (12), where an evidential predicate is embedded under an epistemic one, with no clear logical incongruity: '° (12) a. E’ probabile che sia evidente che lui e’ il colpevole. ‘It is probable that it is evident that he is the guilty one.’

Consider a couple of additional cases. It was noted in chapter 4 that Prospective Aspect AdvPs (almost/imminently) are lower than Retrospective/Proximative Aspect AdvPs (just, soon, etc.). See the contrast in (13):

136 ADVERBS AND FUNCTIONAL HEADS

(13) a. He will soon almost be there. b. *He will almost soon be there.

There 1s, however, no logical ban on having a Proximative adverb embedded under a Prospective predicate such as “to be about to”: (14) He is about to soon be admitted to hospital.

Thus the rigid relative order of the two classes of elements, within the same clause (Proximative > Prospective) cannot be reduced to the logical incongruity of the reverse scope. In the same chapter, we saw that Habitual Aspect precedes Terminative Aspect. This can be seen from the well-formedness of (15a) versus the ill-formedness of (15b), which involve Habitual and Terminative Aspect adverbs: (15) a. Dopo le 10, Gianni non beve di solito pit niente. ‘After 10 o’clock, G. drinks usually no longer anything.’ b. *Dopo le 10, Gianni non beve pit di solito niente.

This order too does not seem to be reducible to a semantic scope restriction, as nothing apparently prevents a Terminative predicate from embedding a Habitual Aspect adverb. See (16): (16) Gianni ha smesso di andare di solito a trovare suo padre la sera. ‘G. stopped usually going to visit his father in the evening.’

All in all, it seems reasonable to conclude that the syntactic order of functional projections cannot be entirely reduced to the semantic scope relations holding among them.!”

6.4 Alleged parametric variation in the relative order of functional heads In the recent literature it is sometimes claimed that the relative order of functional heads is subject to parametric variation across languages. Interestingly, the cases which are brought up to support this conclusion all involve, in one way or another, the position of negation or of agreement with regard to other functional heads, especially Tense (see Laka 1990; Ouhalla 1988, 1990, 1991; Mitchell 1993, among others). I am aware of no claims in the literature to the effect that the relative position of Voice and Tense morphemes, or of Aspect and Tense morphemes, or of Tense and speech act Mood morphemes, is parameterized across languages. The evidence reviewed in chapter 3 also pointed to the conclusion that even within each category (Mood, Tense, Modality, and Aspect) there is a rigid universal order of heads, apparently not open to parametric variation.

Given the evidence reviewed in chapter 5 for the variable (and multiple) positioning of NegPs and DP-related projections, the attested variation concerning negation and agreement should be expected, and, more importantly, should not lead one to the conclusion that parametric variation generalizes to al] functional heads. As noted, it is possible to interpret the different (or multiple) positioning of negation

SOME IMPLICATIONS AND RESIDUAL QUESTIONS 137

and agreement as stemming from a “spell-out” difference: whether a language expresses overtly a lower or higher Agr or Neg (or more than one).'* This might be the case of Turkish and Berber, as discussed in Ouhalla (1990, 189, 192ff; 1991, 136ff): (17) a. Ur-ad-y-xdel Mohand dudsha. (Berber = (9b) of Ouhalla 1990, 189) neg-will-3ms-arrive M. tomorrow *M. will not arrive tomorrow.’ b. Kimsenin geg gelmesini iste-me-di-ler. (Turkish = (9a) of Ouhalla 1991, 57; from Kornfilt 1985) nobody late come want-NEG-PAST-3PIl ‘They did not want anyone to come late.’

According to Ouhalla, the difference between the two languages (whereby negation is inside Tense and Agreement in Turkish, and outside Tense and Agreement in Berber) suggests the parametric difference in underlying structure between the two languages shown in (18): (18) a. LappAgr [rp Tense [yegp Neg Lyp V 11)] (Turkish) b. [Negp Neg [rp Tense [agpAgr [vp V ]]]] (Berber)

This is not particularly surprising, as noted, as the difference involves the position of negation and of agreement which can vary even within the same language.!°

6.5 Hierarchies of nonclausal functional projections A systematic examination of the functional projections embedding other phrases (PPs, NPs, APs, . . .) is out of the question here. The brief survey below, based on the partial results achieved in some of these areas, has the limited goal of rendering it plausible that similarly rigid hierarchies exist in nonclausal phrases as well. Take PPs. Building on Riemsdijk (1978, 1990), Koopman (1993) proposes a quite articulate internal structure for Dutch (Germanic) PPs; one which contains both lexical and functional projections. Simplified in part, her proposal amounts to assuming an extended projection made up of a PP, dominated by a first functional layer (PlaceP),

containing prepositions of static location, dominated by another functional layer (PathP), containing prepositions marking direction or path (see also Jackendoff 1990):7° (19) [ patnp Path® [ piacep Place® [ pp P” DP] ] ]

From this concatenation of head-initial phrases various possibilities found in Dutch are shown to be derivable by head-to-head raising and/or complement-to-spec raising. See (20):7!

Lo

(20) a. (P DP) [ pap Path® [ piacep OP [pp P® [ pp de tafel | ] ] ] (op de tafel) b. (DP P) [pane Path’ [ piacep OP [pp P° [ pp de berg ] ] ] (de berg op) c. (P DP P) [ pap door [ piacep onder [ pp P® [ pp de brug | | ] }] (onder de brug door)

d. (DP PP) [ panp door [ piacep onder [ pp P° [ pp de brug ] | | ] (de brug onderdoor)

138 ADVERBS AND FUNCTIONAL HEADS

Apparently, the same relative scope of Path and Place is found in English, and Romance (where, as a norm, neither heads nor complements raise overtly): (21) [ panp From [ piacep out [ pp of | pp the darkness ] | | | (22) [ PathP Da [ PlaceP dietro [ PP di [ DP noi ] ] ] ]

(lit.) From behind of us

More interestingly, the same relative scope of Path°, Place®, and P®° is found to hold in “head-final” Lezgian if its different “locative Case” suffixes are analyzed as bound Path’, Place®’, and P° morphemes, forcing successive incorporations of the head N (of the DP complement) to yield the mirror image sequence of (23):* (23) N-suffp-suffp,ce-suffpany,

As shown in (24), taken from Riemsdijk (1996, ex. 35), after Haspelmath (1993), three locative Case suffixes are clearly identifiable: the outermost, when present,

indicates direction (either ‘to’ or ‘from’), the middle location (‘at’, ‘behind’, ‘under’, ‘on’, or ‘in’), and the innermost is an “oblique stem marker,” which van Riemsdijk suggests should be considered as the “bound equivalent of [the] dummy preposition[s]” of (of English) and de (of French):

(24) Adessive sew-re-w at the bear

Adelative sew-re-w-d] from the bear

Addirective sew-re-w-di toward the bear

Postessive sew-re-q" behind the bear

Postelative sew-re-q'-aj from behind the bear Postdirective sew-re-q'-di to behind the bear

Subessive sew-re-k under the bear

Subelative sew-re-k-aj from under the bear Subdirective sew-re-k-di to under the bear

Superessive sew-re-l on the bear Superelative sew-re-l-aj off the bear

Superdirective sew-re-l-di onto the bear

Inessive sew-re in the bear

Inelative sew-re-aj (sewrdj) out of the bear Indirective (does not exist) into the bear The structure in (19), which is at the basis of (23) as well, is likely to be a fragment of the full internal structure of PPs.*4 It also seems to be at the basis of the structure

of PPs in the West African Gbe language Gungbe, if, following Aboh (1996), we take the DP (or PP) to raise to a Spec position intermediate between Place” and Path® (25):

(25) a. Asiba zé kwé dé tavo 16 jf. A. put(-PERF) money to table the on “?A. has put money on the table.’ b. mi f3 sSn zan 13 ji. you get-up-IMP from bed the on ‘Get up from the bed!’

SOME IMPLICATIONS AND RESIDUAL QUESTIONS 139

The brief comparison here of West Germanic, Romance, English, Caucasian, and Gbe is sufficient to render the existence of a universal hierarchy of projections within PPs at least plausible. In the case of noun phrases, typological evidence exists that demonstratives are in the Spec of a projection higher than that hosting numerals, which is higher than that hosting descriptive adjectives (see Cinque 1996, §4, and references cited there): ...Lxp Dem X° [ yp Num Y° [ zp Adj Z° .. . NP ] ] ]. In turn, various classes of descriptive adjectives are found to co-occur in a certain order, which remains to be investigated more carefully (for some discussion and references, see Cinque 1994). Universal quantifiers head a projection higher than that hosting Demonstratives (Shlonsky 1991; Giusti 1991, 1993), with genitive arguments-appearing in a number of specific places along the hierarchy. Noun phrases too thus appear to have a rich universal functional structure. The internal structure of Adjective Phrases (and Adverb Phrases) is perhaps the least studied. Although recent work by Zamparelli (1995) and Corver (1997a,b) suggests the presence of an internal relatively rich functional structure for them too, not much comparative work is available. Even this quick survey encourages the conclusion that fixed universal hierarchies of functional projections are likely to be found not only in sentences but also throughout all major phrases.*°

/

Conclusions

The central concern of this study has been the functional structure of the clause. The evidence reviewed here bearing on this question has pointed to the existence of a particularly rich functional make-up of the sentence; one that does not vary across languages. The starting points for this conclusion have been two prima facie unrelated observations. The first was that the various classes of adverbs (AdvPs) enter a rigidly ordered sequence; the same across languages. The possibility (manifested overtly in some languages) of having a head to the immediate right and left of each such AdvP was interpreted as evidence that they occupy the specifier position of distinct phrases:

an empirical conclusion that converges with the purely theoretical conclusion of Kayne (1994, chap. 3, n. 31). The second, crucial, observation was that, if one sets aside agreement and nega-

tion, the order of the head morphemes encoding the different types of functional notions in the clause (mood, modality, tense, aspect, and voice) is also rigidly fixed, and apparently invariant across languages. This was seen to hold whether the head morpheme was a suffix, an auxiliary, or a particle, the order of suffixes being the mirror image of the order of the auxiliaries and particles (of “head-initial” languages). More important still was the subsequent observation that the two hierarchies (that of AdvPs, and that of functional heads) match systematically, from left to right. The transparent semantic relation holding between each adverb class and the contiguous head morpheme to its right (when the two hierarchies are matched) has been taken to suggest that each AdvP is the specifier of the phrase projected by the corresponding functional head morpheme. 140

CONCLUSIONS 141

Even if AdvPs, in certain languages, correspond to particular functional heads that receive no morphological realization, the functional projection was nonetheless taken to be structurally present. This is in line with the restrictive view that language variation reduces to differences in the morphological and lexical realization of the same abstract underlying notions. The indisputable variation that agreement and negation show in their collocation with other functional elements was taken not to weaken the conclusion that UG imposes a rigid hierarchy of functional projections. Agreement and negation stand out as rather special among the other functional elements since they can occur in several distinct positions even within the same language, sometimes simultaneously. It is thus possible to regard their different positioning as a pure morphological effect: whether they are lexically realized in one or the other of the various potential positions they can occupy. More tentatively, [ have also suggested that the same, rich, hierarchy of functional projections (possibly, on the order of 40) 1s present in all languages, and in every sentence of each language, even when no morphological material overtly realizes the corresponding head or specifier. The basic motivation for that assumption resides 1n the intrinsic make-up of functional projections. If each comes with two values, a default and a marked one (the latter typically being realized morphologically more often than the former), then even the simplest sentence of any one language can be taken to contain the entire array of functional projections (with default values). Although many (perhaps, most) of the relative orders among functional elements may ultimately reduce to scope relations among what we can take to be different semantic operators (over the predicate-argument nucleus of the sentence), not all orders are so explicable, it seems. In this case, the hierarchy of functional projections may turn out to be a property of the computational component of UG. There is no need to emphasize the incomplete and provisional character of most of the conclusions reached in this study. Many specific claims will have to be modified; others rejected. Yet, should the hierarchies of AdvPs and of functional heads indeed prove to match systematically, we will have gained a new insight into the structure of UG.

APPENDIX 1

Some Remarks on Other Verbal Forms and Other Romance Varieties

The aim of this appendix is not so much that of pursuing a systematic analysis of the other verbal forms of Italian, or of other Romance varieties, but rather that of providing a wider, comparative, argument for the analysis presented 1n chapter 2 (namely, for the conclusion that word order differences involving verb forms and adverbs is a function of verb movement “around” the adverbs, found in a rigidly fixed sequence). The logic of the argument is the following: the differences involving the position of the same verb form across varieties, or of two verb forms within the same variety, can be stated in a simple and restrictive way if a single hierarchy of AdvPs is assumed, and if the verb 1s taken to raise to one (or more) of the different head positions interspersed in the sequence of AdvPs, with variations depending on the specific verbal form and variety:

(1) ...__ AdvP,; __ AdvP, __ AdvP,;__ AdvP,...V In either case, it is expected that if a verb can be found to the left of a certain AdvP, say AdvP, of (1), then it will necessarily also be found to the left of all lower AdvPs (AdvP; and AdvP, of (1).) An analysis which instead freely adjoined AdvPs to maximal projections to the left and to the right of a verb would make it accidental that if AdvP, (of (1)) can only be adjoined to the left of the verb (and not to its right), so must AdvP,. In such an analysis, the implication just mentioned 1s not naturally derivable, as the relative order of the verb and the adverbs does not interact with the relative order of the AdvPs among each other. Subjecting various verbal categories to this probe will also reveal certain patterns which have gone unnoticed in the literature and which bear on the proper analysis of the relevant constructions. 142

SOME REMARKS ON VERBAL FORMS AND OTHER ROMANCE VARIETIES 143

The level of inquiry will necessarily be descriptive. We have not yet determined precisely the nature of the heads in between the various AdvPs (but even with the insight of the previous chapters, the question of what triggers the movement to a particular position still remains to be understood).

A.1 Infinitives in French and Italian As pointed out in Pollock (1989, 412) and Belletti (1990, 70ff) (lexical) infinitives in Italian raise more than the corresponding infinitives in French, resembling finite verbs. Focusing on their relative order with respect to various classes of AdvPs, we may ask to which positions infinitives, in the two languages, raise obligatorily, and to which they raise in an apparently optional way. From Engver’s (1972) careful study on adverb/infinitive order in (literary)

French, and from Pollock (1989), one can draw the following conclusions for French: 1. Infinitives apparently do not have to move obligatorily to any higher functional head, as they can follow all the AdvPs isolated in §1.2, just like past participles (see (2) and (4)--(5)).! 2. Differently from past participles, they can optionally raise to various heads, up to the head between plus and toujours, but not any higher (see (3) and (4)):?

(2) a. Il faut trés bien le loger. ‘It is necessary to stock it very well.’ (Engver 1972, 74) b. la liberté de penser, et de mal penser ‘the freedom to think and to think badly’ (Engver 1972, 74) c. Tu a du beaucoup souffrir. “You had to suffer a lot.’ (Engver 1972, 88) d. J’ai recommandé ... de ne guére s’arréter. ‘recommended ... not to stop.’ (Engver 1972, 28) e. Je pensais ne jamais te revoir. ‘I thought I would never see you again.’ (Engver 1972, 25) f. Ik faut toujours le loger. ‘It is necessary always to stock it.” (Engver 1972, 75)

(3) a. Il faut le loger bien. ‘It is necessary to stock it well.’ (Engver 1972, 74) b. le temps de répondre e de répondre mal. ‘the time to answer, and to answer badly.’ (Engver 1972, 98) c. Il a du souffrir beaucoup. ‘He had to suffer a lot.’ (Engver 1972, 88) d. Car ce objet parait ne vous préoccuper guére. “Because this object seems not to worry you.’ (Engver 1972, 28) e. J’ai cru n’arriver jamais. ‘[ thought I would never arrive.’ (Engver 1972, 26) ‘Tl faut le loger toujours. ‘It is necessary always to stock it.’ (Engver 1972, 75)

144 APPENDIX 1

(4) a. Je pensais ne plus te revoir / *ne te revoir plus. ‘I thought I would no longer see you / see you no longer.’ b. Il voulait ne pas partir / *ne partir pas. ‘He wanted not to leave / to leave not.’

Even if Engver (1972) does not provide examples, complétement and tout can also be found either preceding or following an infinitive; as expected, given that they come between toujours and bien. See (5) and (6): (5) a. Il faut complétement le refaire. ‘One must completely do it again.’ b. Il faut le refaire complétement. ‘One must do it again completely.’

(6) a. II faut tout faire trés bien. “One must everything do very well.’ b. Il faut faire tout trés bien. ‘One must do everything very well.’ Concerning Italian, we mentioned Belletti’s (1990, 70ff) observation that infinitives in Italian appear to raise as high as finite verbs. Considering their distribution with a number of adverbs in infinitival relatives (pp. 75ff), she notes a complete parallelism with the movement of finite verbs. Just like finite verbs, they appear to necessarily raise past “lower” adverbs ((7a), her (107b)), but to be able to remain to the right of sentential adverbs ((7b), her (106a)), respectively):? (7) a. *?Ho trovato qualcuno a cut spesso affidare questo tipo di incarico: ...a cul affidare spesso) ‘I found someone to whom often to assign this kind of task.’ b. Cerco un uomo al quale possibilmente/forse presentare Maria. ‘!’m looking for a man to whom if possible/perhaps to introduce Maria.’

If we extend our examination to other infinitival contexts and to the rest of the “higher” adverbs of §1.2) an interesting contrast emerges. To begin with, while infinitives in wh-constructions behave essentially as finite verbs in their obligatory and optional movements, as Belletti observed, infinitives in subject and complement clauses appear to raise obligatorily to the highest head which finite verbs raise to only optionally or cannot even raise to. So, the infinitive in such clauses turns out to obligatorily precede “speech act” AdvPs and all AdvPs following in the hierarchy), whereas finite verbs raise to the left of bene ‘well’ obligatorily, but to the left of all higher AdvPs

only optionally.4 Compare (i) of n. 7 in chapter 5 and (15)-(17) of chapter 2 with (8)—-(9) here:

(8) a. Mi mette in imbarazzo, (*francamente) disporre (francamente) di mezzi cosi limitati. ‘It embarrases me frankly to have so limited means.’ b. Lo salvo, (*per fortuna) godere (per fortuna) del loro appoggio. ‘It saved him (fortunately) to avail himself (fortunately) of their support.’ c. Lo lusingava, (*probabilmente) conservare (probabilmente) la loro stima. ‘It flattered him (probably) to maintain (probably) their esteem.’ d. Lo elettrizzava, (*ora) prendere (ora) liniziativa.° ‘It excited him (now) to take (now) the initiative.’ e. Non lo entusiasmava, (*forse) ottenere (forse) la loro fiducia. ‘It did not arouse his enthusiasm (perhaps) to obtain (perhaps) their trust.’

SOME REMARKS ON VERBAL FORMS AND OTHER ROMANCE VARIETIES 145

f. (*Intelligentemente) rifiutarsi (intelligentemente) di firmare 1o ha salvato. ‘(Intelligently) to refuse (intelligently) to sign saved him.’

(9) a. (*Francamente) esser (francamente) meno ricco, e meno importunato, vorrei! ‘(Frankly) to be (frankly) less rich, and less bothered, I would like!’ b. (*Purtroppo) essere (purtroppo) arrivato tra gli ultimi, detestava! ‘(Unfortunately) to have (unfortunately) arrived among the last, he hated!’ c. (*Probabilmente) essere (probabilmente) il migliore, avrebbe voluto! ‘(Probably) to be (probably) the best, he would have liked.’ d. (*Ora) avere (ora) il loro appoggio, avrei desiderato! ‘(Now) to have (now) their support, I would have liked!’ e. (*Forse) ottenere (forse) un risultato migliore, avrei preferito! ‘(Perhaps) to obtain (perhaps) a better result, I would have preferred!’ f. (*Astutamente) essermi (astutamente) rifiutato di firmare, vorrei!® ‘(Smartly) to have (smartly) refused to sign, I'd like!’

Notice that it cannot be claimed that the infinitive precedes all “higher” AdvPs because it moves to C°. Subject infinitives, and complement infinitives, in Italian (except for those under verbs of thinking and saying; Rizzi 1981, 1982, 80) do not allow for Aux-to-COMP: (10) *Esser Gianni disposto a restare ci meraviglio. ‘Tt surprised us that G was willing to stay.’ (11) *Vorrei/preferirei/detesto/occorre/etc. esser Gianni disposto a restare. ‘I would like / would prefer / hate / it is necessary / to be G. willing to stay.’

The head to which the infinitive raises in (8) and (9) has thus to be in the complement of C°.’ Infinitival wh-interrogatives pattern with infinitival relatives in allowing the infinitive to stay to the right of “higher” AdvPs: (12) a. Non so di chi (francamente) potermi (francamente) fidare. ‘I do not know whom (frankly) to be able (frankly) to trust.’ b. Credo di sapere a chi (per fortuna) poter (per fortuna) affidare 1 miei risparmi. ‘T think I know to whom (luckily) to be able (luckily) to entrust my savings.’ c. Si chiedeva di chi (probabilmente) potersi (probabilmente) maggiormente fidare. ‘He wondered whom (probably) to be able (probably) to trust more.’ d. Non so da chi (ora) farmi (ora) aiutare. ‘I don’t know by whom (now) to get (now) myself helped.’ e. Mi chiedo perché (forse) non dimostrargli (forse) pitt fiducia. ‘I wonder why (perhaps) not to show him (perhaps) more trust.’ f. Non so se (intelligentemente) rifiutarmi (intelligentemente) di firmare, o no. ‘I don’t know if (intelligently) to refuse (intelligently) to sign, or not.’

The presence of a focalized or (clitic) left dislocated phrase at the front of an infinitival complement sentence also permits the infinitive to stay lower (perhaps, with some degree of marginality):° (13) a. Avrei preferito A GIANNI / a Gianni (?francamente) non dover (francamente) affidare anche questo compito. ‘I would have preferred to G. (focus) / (topic) (frankly) not to have (frankly) assigned this task too.’

146 APPENDIX 1 b. Occorre A LORO / a loro (?purtroppo) affidare (purtroppo) anche questo compito delicato. ‘It is necessary to them (focus)/(topic) (unfortunately) to (unfortunately) assign this delicate task too.’ c. Vorrei A LORO / a loro (?probabilmente) non dover (probabilmente) rispondere delle mie azioni. ‘I would like to them (focus) / (topic) (probably) not to have (probably) to account for my acts.’ d. Avrei preferito A GIANNI / a Gianni (?allora) poter (allora) affidare l’incarico. ‘I would have preferred to G. (focus) / to G. (topic) to be able to assign this task.’ e. Avrei preferito A GIANNI / a Gianni (?forse) poter (forse) affidare anche questo incarico. ‘. would have preferred to G. (focus) / to G. (topic) (perhaps) to be able (perhaps) to assign this task as well.’ f. Avrei preferito A GIANNI / a Gianni (?intelligentemente) aver (intelligentemente) affidato tutti 1 miei risparmi. ‘I would have preferred to G. (focus) / to G. (topic) (intelligently) to have (intelligently) entrusted all of my savings.’

Finally, it was noted in Cinque (1983) that (in all styles) some “higher” adverbs can intervene between an infinitive and a real preposition taking it as a complement, differently from the case of prepositional complementizers. See, for example, (14)—(15): (14) a. Gli hanno parlato pitt volte senza forse riuscire a convincerlo. ‘They talked to him a few times without perhaps managing to convince him.’ b. *Sono riuscito a forse convincerlo. ‘I managed to perhaps convince him.’

(15) a. Sié voluto dimettere per mai pit dover subire ingiurie simili. ‘He decided to resign never to have to put up with such affronts again.’ b. *Cercherod di mai pitt rivedere quell’ individuo. ‘I will try never to see that person again.’

A.2. Past participles in some Romance varieties As noted in §2.1, active past participles raise obligatorily to distinct heads in different Romance varieties; hence are (obligatorily) found to the left of different classes of adverbs in a way that is consistent with the fixed order of adverbs seen above. While in French, as seen, an active past participle does not need to (cannot) raise past bien ‘well’ (only past tét ‘early’, and few other adverbs; Togeby 1984, 208), in most other varieties it must. In Sardinian, it raises past bene but need not raise any higher. In Italian, instead, it must raise higher than tutto ‘everything’ (and bene ‘well’ and presto ‘early’). In Bellunese, it must apparently raise obligatorily to the left of del tut ‘completely’, and in Catalan higher still.° Various (northern Italian) Romance varieties appear to differ also with respect to the heads to which the (active) past participle raises “optionally.” See (16), which summarizes the findings relative to some such varieties:!”

SOME REMARKS ON VERBAL FORMS AND OTHER ROMANCE VARIETIES 147

(16)

Italian: V mica V gia V pit V sempre . completamente V tutto * bene * Friulian: V mingul V za V plui V simpri ¥ completaminti V dut * ben * Venetian: * miga * gia 7? pit ? sempre V del tuto V tuto * ben * Trevisan: *mia * za V pit: V sempre V del tuto 2? tuto * ben * Paduan: * miga V za V piu V sempre V del tuto * tuto * ben * Bellunese: * mia * za * pi 2? sempre V del tut * tuto * ben * Pavese: V mia V giamo V pi V sempar V no V dal tut * tut (cos) * ben * Milanese: * minga V gemo V pii V semper V no * dal tiit * tiiscds/tiit * ben * Piedmontese: *pa * gia * pi nen V sempre V dal tiit V tiit * ben *

Passive past participles appear to be able to raise lower than active ones (under certain circumstances). In Italian, for example, while, as noted, active past participles cannot be to the right of tutto and bene (see (17a~c)), passive ones can, in generic contexts (see (18a-—c)):

(17) a. *Ho tutto bene arrangiato. ‘T have everything well arranged.’ b. *?Ho tutto arrangiato bene. ‘[ have everything arranged well.’ c. Ho arrangiato tutto bene. ‘I have arranged everything well.’ (18) a. (?) Per fortuna, é stato tutto bene arrangiato. ‘Luckily, has been everything well arranged.’ b. Per fortuna, é stato tutto arrangiato bene. “Luckily, has been everything arranged well.’ c. Per fortuna, é stato arrangiato tutto bene. ‘Luckily, has been arranged everything well.’

Apparently similar facts hold in Romanian (see (19)-(20); Carmen Dobrovie-Sorin, personal communication) and Spanish (see (21)—-(23); Lois 1989, 34, 40). While active

past participles cannot be separated by adverbs from finite auxiliaries, passive past participles can:!! (19) a. *Am tot organizat. ‘I have everything organized.’ b. *N-am niciodata citit cartea. ‘I have never read the book.’ (20) a. A fost tot organizat. ‘Everything has been organized.’ b. Cartea n-a fost niciodata citita. | ‘The book has been never read.’ (21) *Juan (no) ha/habia nunca/evidentemente/siempre/bien/pues/todavia/ya visto a Maria. ‘J. has/had (not) ever/evidently/always/etc. seen M.’ versus (22) Juan no ha/habia visto nunca/evidentemente/ siempre/ pues/ todavia/ ya a Marfa.

(23) Los estudiantes (no) | fueron nunca arrestados (por la policia). han sido siempre

pues : desgraciadamente

todavia ya

“The students (not) were ever/always/etc. arrested (by the police.)’

148 APPENDIX 1

The contrast between (22) and (23) might be interpreted as evidence that passive past participles raise less than active ones, but this is unclear. Alongside the order in (22), given in Lois (1989, 34), in which the active past participle is found to the left of the various adverb classes, the order in (24) also appears possible in which the auxiliary and the adjacent past participle have apparently remained lower, to the right of the same classes of AdvPs: (24) Juan nunca/evidentemente/siempre/bien/pues/todavia/ya ha/habia visto a Maria.

This may suggest that the contrast is due to some other property distinguishing the two types of participles, one by which active, but not passive, past participles must be adjacent to the finite auxiliary.

A.3. Absolute past participles, present participles, and gerunds in Italian Absolute past participles in Italian appear to move to distinct positions depending on whether they enter a control construction (25a) or a nominative subject construction (25b). For the latter, there 1s evidence that they move to C° (see Kayne 1989a; Cingue 1995, 231f n. 24; Belletti 1990): (25) a. Arrivato puntuale, Gianni si indispetti del nostro ritardo. ‘Having arrived punctual, G. got annoyed at our delay.’ b. Arrivato puntuale anche Gianni, potemmo partire subito. ‘Even G. having arrived punctual, we could leave immediately.’

We should thus expect the latter to precede all AdvPs, even the highest, which is what we find:!? (26) a. Arrivato fortunatamente anche Gianni, potemmo partire subito. ‘Even G. having luckily arrived, we could leave immediately.’ b. *Fortunatamente arrivato anche Gianni, potemmo partire subito. ‘Luckily even G. having arrived, we could leave immediately.’ c. Arrivato anche Gianni fortunatamente *(in tempo), potemmo partire subito. ‘Even G. having arrived luckily (in time), we could leave immediately.’ (27) a. Arrivato evidentemente anche Gianni, hanno deciso di partire. ‘Even G. having evidently arrived, they decided to leave.’ b. *Evidentemente arrivato anche Gianni, hanno deciso di partire. ‘Evidently even G. having arrived, they decided to leave.’ c. Arrivato anche Gianni evidentemente *(in tempo), hanno deciso di partire. ‘Even G. having arrived evidently (in time), they decided to leave.’

(28) a. Arrivato allora anche Gianni, potemmo partire alle cinque. ‘Even G, having then arrived, we could leave at five.’ b. *Allora arrivato anche Gianni, potemmo partire alle cinque. “Then even G, having arrived, we could leave at five.’ c. Arrivato anche Gianni allora, potemmo partire alle cinque. ‘Even G. having arrived then, we could leave at five.’

SOME REMARKS ON VERBAL FORMS AND OTHER ROMANCE VARIETIES 149

(29) a. Arrivato forse anche Gianni, Maria decise di restare. ‘Even G. having perhaps arrived, M. decided to stay.’ b. *Forse arrivato anche Gianni, Maria decise di restare. ‘Perhaps even G. having arrived, M. decided to stay.’ c. Arrivato anche Gianni forse *(in ritardo), Maria decise di restare. ‘Even G. having arrived perhaps (late), M. decided to stay.’ (30) a. Arrivato intelligentemente anche Gianni, poterono aiutarlo. “Even G. having intelligently arrived, they could help him.’ b. *Intelligentemente arrivato anche Gianni, poterono aiutarlo. ‘Intelligently even G. having arrived, they could help him.’ c. Arrivato anche Gianni intelligentemente *(puntuale), poterono aiutarlo. ‘Even G. having arrived intelligently (punctual), they could help him.’

As expected, the past participle in the Aux-to-COMP absolute construction also precedes all lower AdvPs (I omit relevant examples).

In the control construction, instead, the past participle can follow (as well as precede) all “higher” AdvPs (of §1.2), which is compatible with it not necessarily moving to C. See:!° (31) a. Fortunatamente arrivato in anticipo, Gianni pote rimediare. b. Arrivato fortunatamente in anticipo, Gianni poté rimediare. ‘Having luckily arrived early, G. could remedy.’ (32) a. Evidentemente attratto dalla prospettiva del guadagno, Gianni accetto. b. Attratto evidentemente dalla prospettiva del guadagno, Gianni accetto. ‘Being attracted by the prospect of earning money, G. accepted.’

(33) a. Probabilmente arrivato in anticipo, Gianni se ne riando. b. Arrivato probabilmente in anticipo, Gianni se ne riando. ‘Having probably arrived early, G. went away again.’ (34) a. Anche allora arrivato in anticipo, Gianni si sedette. b. Arrivato anche allora in anticipo, Gianni si sedette. ‘Having even then arrived early, G. sat down.’ (35) a. Forse arrivato in anticipo, Gianni se ne riando. b. Arrivato forse in anticipo, Gianni se ne riando. ‘Having perhaps arrived early, G. went away again.’ (36) a. Saggiamente arrivato in anticipo, Gianni poté rimediare. b. Arrivato saggiamente in anticipo, Gianni poté rimediare. ‘Having wisely arrived early, G. could remedy.’

If we consider the “lower” pre-VP AdvPs of §1.1, we can observe the same “cutting point” characterizing active past participles. Absolute past participles can be found to either precede, or follow, habitual adverbs, negative adverbs, gid, pit, sempre and completamente, but they have to precede bene. (37) a. Solitamente arrivato sempre puntuale, Gianni da quella voita arrivo in ritardo. b. Arrivato solitamente sempre puntuale, Gianni da quella volta arrivé in ritardo. ‘Having usually arrived always punctual, G. since then arrived late.’

150 APPENDIX 1

(38) a. Non affatto arrivato sempre puntuale, Gianni da quella volta arrivd tardissimo.'4 b. Arrivato non affatto sempre puntuale, Gianni da quella volta arrivo tardissimo. ‘Having not at all always arrived punctual, G. since then arrived very late.’

(39) a. Gia arrivato verso le sei, Gianni decise di restare. b. Arrivato gia verso le sei, Gianni decise di restare. ‘Having already arrived towards six o’clock, G. decided to stay.’

(40) a. Non piu arrivato in ritardo, Gianni, da allora, fu apprezzato da tutti. b. Arrivato non piz in ritardo, Gianni, da allora, fu apprezzato da tutti. ‘Having no longer arrived late, G., since then, was appreciated by everybody.’

(41) a. Sempre arrivato puntuale, Gianni quella volta sorprese tutti.' b. Arrivato sempre puntuale, Gianni quella volta sorprese tutti. ‘Having always arrived punctual, G. that time surprised everybody.’

(42) a. (Pur) completamente bruciata, 1a torta fece la sua bella figura. b. (Pur) bruciata completamente, \a torta fece la sua bella figura. ‘(Although) being completely burned, the cake made a good impression.’

(43) a. *Bene arrivata, la cassa si apri senza difficolta. b. Arrivata bene, la cassa si apri senza difficolta. ‘Having arrived well, the chest opened with no difficulty.’

Similar facts are found with present participles and gerunds. See (44)—(55) and (56)— (59):16

(44) a. Gli aspetti fortunatamente riguardanti il nostro caso. b. Gli aspetti riguardanti fortunatamente il nostro caso. ‘The aspects luckily concerning our case.’

(45) a. Gli aspetti evidentemente riguardantt il primo dei due casi. b. Gli aspetti riguardanti evidentemente il primo dei due casi. ‘The aspects evidently concerning the first of the two cases.’ (46) a. Le disposizioni probabilmente concernentici pit! da vicino. b. Le disposizioni concernentici probabilmente pit da vicino. ‘The dispositions probably concerning us most closely.’

(47) a. L’edificio allora sovrastante la piazza del paese. b. L’edificto sovrastante allora la piazza del paese. ‘The building then dominating the village square.’

(48) a. Il ritratto forse raffigurante la compagna dell’ artista. b. I] ritratto raffigurante forse la compagna dell’ artista. ‘The painting perhaps representing the artist’s partner.’ (49) a. I pannelli stupidamente simboleggianti il potere del regime. b. I pannelli simboleggianti stupidamente il potere del regime. ‘The panels stupidly symbolizing the government power.’ (50) a. Gli argomenti normalmente riproducenti 1 risultati voluti. b. Gh argomenti riproducenti normalmente 1 risultati voluti. ‘The arguments normally reproducing the desired results.’

SOME REMARKS ON VERBAL FORMS AND OTHER ROMANCE VARIETIES 151

(51) a. I documenti gia riguardanti il nostro caso. b. I documenti riguardanti gia il nostro caso. ‘The documents already concerning our case.’

(52) a. Gli argomenti non pit dirimenti le questioni di fondo. b. Gli argomenti non dirimenti pit le questioni di fondo. ‘The arguments no longer settling the fundamental questions.’ (53) a. ?I valori sempre determinanti lo stesso risultato. b. I valori determinanti sempre lo stesso risultato. ‘The values always determining the same result.’

(54) a. ?Gli unici risultati completamente inficianti l’argomento. b. Gli unici risultati inficianti completamente |’ argomento. ‘The results completely impugning the argument.’ (55) a. *Le descrizioni bene riproducenti Il’esperimento. b. Le descrizioni riproducenti bene l’esperimento. ‘The descriptions well reproducing the experiment.’

(56) a. Mene andai | francamente non sospettando nulla. fortunatamente evidentemente probabilmente

b. Me ne andai, non sospettando | francamente nulla. fortunatamente evidentemente probabilmente ‘T left frankly/luckily/evidently/probably not suspecting anything.’

(57) a. Sene andava, | allora non sospettando nulla. forse stupidamente solitamente

b. Se ne andava, non sospettando | allora nulla. forse stupidamente solitamente

‘He used to leave then/perhaps/stupidly/usually not suspecting anything.’

(58) a. Cominciava [ gia travisando Io spirito dell’ iniziativa. non pill sempre completamente

b. Cominciava travisando | gia lo spirito dell’ iniziativa. non piu sempre completamente

152 APPENDIX 1 ‘He used to start already/no longer/always/completely distorting the spirit of the initiative.’

(59) a. *Cominciava bene interpretando lo spirito dell’ iniziativa. b. Cominciava interpretando bene lo spirito dell’ iniziativa. ‘He used to start interpreting well the spirit of the initiative.’

A.4_ Finite verbs in some Romance varieties In the unmarked case, finite verbs in Italian precede mica and all adverbs following mica (gid, piu, ancora, sempre, appena, subito, brevemente, quasi, completamente, bene, presto), while preceding or following all higher adverbs (rapidamente, spesso, di nuovo, di solito, stupidamente, forse, obbligatoriamente, volentieri, necessariamente, ora, probabilmente, evidentemente, fortunatamente, francamente).'" In this they appear to contrast with finite verbs in Romanian and Spanish, which typically follow a number of “lower” adverbs (see (60)—(62)). But this difference remains to be better investigated: (60) a. Nu mai cred cae posibil. (Dobrovie-Sorin 1994, 10) b. *Nu cred mai ca e posibil. ‘TY no longer think that it is possible.’

(61) a. Prea poate spune ce vrea. (Dobrovie-Sorin 1994, 26) b. *Poate prea spune ce vrea. ‘(He) can too much say what (he) wants.’

(62) a. Juan a menudo/rara vez esta deprimido. (Lois 1989, 23) ‘J. often/rarely is depressed.’ b. El nifio casi/apenas camina. ‘The child almost/just walks.’ c. Juan en seguida/immediatamente piensa que nadie lo quiere. ‘J. soon/immediately thinks that no one loves him.’ d. Juan ya llego. ‘J. already has arrived.’

APPENDIX 2

A Synopsis of the Orders of Overt Functional Heads in Individual Languages

In this appendix, the relative orders of functional heads in a number of languages are given for which overt morphological evidence exists within the language. Whenever there is some work explicitly proposing a particular order of functional heads, I simply quote that work in parentheses, referring to it for the relevant evidence. Where (to my knowledge) no explicit proposal exists, evidence for the order suggested is given, with an indication of its source. As in the text, [ignore the positions of AgrPs and NegPs. Whenever it was sufficiently clear which particular type of aspect was

involved, I have indicated it (also signaling where I adapted the terminology for uniformity reasons). Needless to say, the danger of misinterpreting the facts of unfamiliar languages remains high. Virtually all of the world’s major language groups happen to be represented here (the classification is based on Ruhlen’s 1991). This should avoid the risk of basing conclusions about the universal order of functional heads only on languages which are genetically related. Even though the works and the sources referred to here are of uneven depth and coverage, the survey will have served its purpose if it reveals, or supports, the existence of at least some generalizations. INDOEUROPEAN

English: T/Mod Asp perfect ASPprogressive Voice V (Chomsky 1957) Spanish: T ASPperfect ASPprogressive VOICe V (Zagona 1988, chap. 5)

Welsh (Celtic): T Aspypertect ASPprogressive V ($3.4 and Ouhalla 1988; Hendrick 1991) 153

154 APPENDIX 2

Modern Greek: T Asp Voice V (Rivero 1990) Hindi: T(Past) T(Anterior) /ASpprogressive YW (see (23a—b) of chapter 3)

FINNO-UGRIC

Finnish: TIMood Gr reatis ASPperfect Voice V (Mitchell 1991; Holmberg et al. 1993) Hungarian: Mood gyeech act 1 (Past) Modo, V

(1) a. Meg-javit-t-ott-ad-e? (Istvan Kenesei, personal communication) COMPL-repair-PAST-2sg-Q ‘Did you repair it?’ b. Janos meg-javit-hat-ta az 6ra-t. (de Groot 1995, 41) J. COMPL-repair-MOD-PAST.3sg the clock-ACC ‘J. could repair the clock.’

ALTAIC

Evenkt: Moodgpecch act T(Present)/T(Past) Asppanituat ASP(imyperfect VOICE / ASPouickty ASP semelfactive / ASPresultative Vv

(2) a. Nuynan eme-che-n-nu, e-che-n-nu? (Nedyalkov 1994, 2) 3sg come-PAST-3sg-Q, Neg-PAST-3sg-Q ‘Did (s)he come, or not?’ b. D’apka e-che-n iche-v-d’e-re. (Nedyaikov 1994, 12) shore Neg-PAST-3sg see-PASS-IMPERF-fixed form of the lexical verb ‘The shore is/was not seen.’ c. ... tuksa-s-malcha-ra-n. (Nedyalkov 1997, 252) ... mun-SEMEL-QUICKLY-PAST-3sg ‘,.. and set out running quickly.’ d. Urke mo:-t somi-cha-v-d’ara-n. (Nedyalkov 1997, 234) door tree-INSTR close-RESULT-PASS-PRES-3s¢ ‘The door is closed with a stick.’ e. Nungan ulle-ve tulile lo: van-d’e-ngne-re-n. (Nedyalkov 1997, 256) she meat-ACC outdoors hang-IMPERF-HAB-PAST-3sg ‘She usually hangs meat outdoors for some time (for drying).’ Korean: Mood speech act MOOdeyajuative MOO yidentiaa MOdepistemic 1(Past) T(Future) Mod

ASPhabituat/ ASPprogressive WOICe V (See §3.1,$3.4 and references cited there)

(Khalkha) Mongolian: Mood gyeech act T(Past) T(Future) Aspprogressive Voice V (see

Svantesson 1991, 191ff) (3) a. ta: mini ax-tal w:lz-son-@ you my brother-COMITAT meet-PAST-Q ‘Have you met my brother?’ b, ...9ox’in-d belag aw-ax bai-son. girl-DAT present buy-FUT(irr) be-PAST ‘... he would have bought a present for the girl.’

OVERT FUNCTIONAL HEADS IN INDIVIDUAL LANGUAGES 155

c. bi: €@l@ aw-¢ moGoi-g Ewl@d-dog bai-la/san. I stone take-PROG snake-ACC throw-HAB be-PAST ‘I used to take a stone and throw it at the snake.’

d. ter jaw-tix-say bai-la/say.! he go-INT-ANT be-PAST ‘He had left.’ e. xan al-ogd-Cix-Ze. king kill-PASS-INT-PAST “The king has been killed.’ Turkish: Mood peech act MOOG gyidentiat MOO epistemic / T(Past) T(Future) Mod T(Anterior)

OF ASPpertect ASPprogressive Voice V (see exs.(5), (66), (74a), (78a), and (i) of n.43 of

chapter 3, and (4) of n.2 of chapter 4) CAUCASIAN

Abkhaz: Mood ,.yidentia / T(Past) T(Puture) / T(Anterior) Aspgequentative V (see Hewitt

1979, 89ff; 1989, 183)

(4) a. yo-q’a-l-c’a-x’a-zaap’. it-PREV-she-do-ANT-EVID ‘Apparently she has done it.’ b. ya- q’a-l-c’a-x’a-zaaran. it-PREV-she-do-ANT-PAST.EVID ‘Apparently she had done it.’ c. ya- q’a-l-c’a-ro-n. it-PREV-she-do-FUT-PAST ‘She would {perhaps} (have) do(ne) it.’ d. ya-z-ba-ka-x’é-yt’. it-I-see-FREQ-ANT-PRES ‘I have already seen it several times.’

Ubykh: T(Future) Mod,,.o: V (5) §’a-k’a-je-fa-na-mo-t. (see Campbell 1991, 1402) we-go-REFL-POTENTIAL-NUM-NEG-FUT “We shall not be able to go back.’ Lezg ian: Mood. yidential / T(Past) T(Future) ASP continuative / ASP progressive Vv

(6) a. Qe sobranie Ze-da-lda. (see Haspelmath 1993, 140ff) today meeting be-FUT-EVID ‘They say that there will be a meeting today.’ b. Im xizan.di-z jeke kitimek Ze-da-j. this-ABS family-DAT big help be-FUT-PAST ‘This would be a big help for the family.’ c. Za-z wué-da-t’a Ci-zma-é-ir. I-DAT [what do-FUT-COND] know-CONT-NEG-PAST? ‘I did not know anymore what to do.’

156 APPENDIX 2 d. Marf dat’ana “qwa-zwa-}. rain constantly fall-PROG-PAST ‘It was raining incessantly.’ AFROASIATIC

Arabic (Semitic): Mod Moodgrreais L/ASp/Voice V (Fassi Fehri 1993, 82ff) M of u-Gudur (Chadic): Mood peech act TP ast) T(Anterior) ASP perfect ASP progressive ASPprospectve W° (See Hollingsworth 1991)

(7) a. Ngaa kwana ya sawa na, ndaw mayal fa da 14 dola daw. Of yesterday I come TOP, man thief PROG PROSP take money my “Yesterday, when I came, a thief was just about to steal my money.’ b. Mahura, ta goley cay. Big PAST grow ANT ‘(He was) big; he had already grown up.’ c. Akorey malagway 1a. She grind corn PERF ‘She will grind corn.’ d. Ka gwaw nga layawa 14 may daw? (p. 245) You can for take4me“DIR PERF also Q ‘Can you gather it up for me too?’

[NILO-SAHARAN] NIGER-CONGO

Dagaare: T(Past) Aspnapituat ASPprogressive VW (Bodomo 1993) Fula/Fulfulde: ASpceteratuve ASPcompletive ASP (semelrepenive (See Arnott 1970; Fagerli 1994)

Gung be: T(Future) ASP habitual ASP progressive ASP prospective V (see §3.3 and Aboh 1993,

1996) Ise Kirt: T(Future) ASPhabitual ASP perfect ASP progressive Vv’

Kako: T(Future) T(Anterior)/Aspeontinuative/ ASPprogressive/ ASPcomplenvey (U. Ermst 1991)?

(8) a. Ata bé ndi ké pesd mé-jati. He FUT be CONT PROG cut c6-tree “He will always be cutting the tree.’ b. A ma si lénsa lambd. He ANT COMPL forget lamp ‘He has completely forgot the lamp.’ Kom: T(Past) ASP pertect ASPpropressive ASPinceptive VW (Chia 1976, 73ff)°

(9) a. Ivui tina st suria. rain PAST, IMPERF INCEPT fall ‘Rain was beginning to fall.’ b. Johnson ti na chén—chén. J. PAST, IMPERF dance-PROG ‘J. was dancing.’

OVERT FUNCTIONAL HEADS IN INDIVIDUAL LANGUAGES 157

Makaa: T(Past) Aspfrequentative ASPprogressve V (See Heath 1991, 4, 12)

(10) Ms 4d ngs dad idtiw ki nd baagé. I PAST FREQ PROG eat food NEG with hot-pepper ‘Il often ate food without hot pepper.’

Northern Sotho. T(Past) T(Future) T(Anterior) Asppropressive V (see §3.6, and n. 55, based on Louwrens, Kotsch and Kotzé 1995, 52ff) Yoruba: Mood gpeech act/T (Anterior) ASPprogressive V (see Adéwole 1989, 1991; Awobultyi 1967, 235ff; Oke 1972)

(11) a. Moti filo. (Oke 1972, 152) I ANT PROG go ‘[ have been going.’ b. sé 0 rf bd? (Adéwolé 1991, 9) Q you PROG come ‘Are you coming?’

[KHOISAN] SINO-TIBETAN

Burmese: Moodgpeech act | Mod V

(12) cano di-lau’ pei: dhin dha la:? (Goral 1988, 16) I this approximately pay suitable non-FUT Q particle ‘Should [ have paid that much?’ Chinese: Mod ASP perfect ASPcompletive V (see §3.1, based on Smith 1987, 1991) Garo: Mod, pistemic T(Past) ASPterminative Vv

(13) a. anti-ci re’an-aha-kon. (Bybee 1985, 180) market-to go-PAST-PROBABLE ‘I think he went to the market.’ b. ca’-man-jok. (Campbell 1991, 487) eat-TERMIN-PAST ‘He has fimished eating.’

Kachin: T(Future) Mod,,,, V (see (56) of chapter 3) Patani: T(Past) Aspprogressive V (See Saxena 1995, 6)

(14)... krapi léki t6-i. cry-nominal PROG be-PAST/3sg *... She was crying.’

Tshangla: T(Future) T(Anterior) Asppyooressive V (see Andvik 1993, 89)

(15) got-chho-wa-uphe. look-PROG-ANT-FUT ‘(He) will have been looking.’

158 APPENDIX 2

[(CHUKCHI-KAMCHATKAN] ESKIMO-ALEUT

Aleut: Mood peech act T(Present) Mood, yidentiat T(Past)/T (Future) ASPhabitual ASPinceptive

Voice V (see Bergsland 1994, 351, 340, 337) (16) a. Hama-n galgada-X taxsa-ga-X agun-un angagsina-z-iin chisi-lga-qali-za-qa-X a-xta-ku-x. that food-sg store-PAST-sg be-COND-3PAST.sg person-PL-to-them distributePASS-INCEPT-HAB-PAST-sg be-EVID-PRES-sg “When that food had been stored it was distributed to the people (it is said).’ b. Saga-tu-laga-agila-a—txin. sleep-much-NEG-FUT-OPT-you ‘Don’t sleep late.’ c. Qila-ga-n ukuxta-duu(ka)-ku-ng. tomorrow see-FUT-PRES-isg ‘Pll see him tomorrow.’ Central Alaskan Yup’ ik: Mood reais ASPpabitual &SPdurative ¥ (See Mithun and Ali 1996)

(17) ayag-uma-lar-tu-q. go-LONG TIME-HAB-INTR.INDIC-3s¢ ‘He customarily goes for long periods of time.’ DRAVIDIAN

Malayalam: T(Present) T(Past) T(Future) Asppertect /ASPprogressive VOICe Causative V

(see Babu 1996, 4ff)’ (18) a. flaan innale aviDe pook-um-aayir-unnu. I-NOM yesterday there go-FUT-be.PAST-PRES ‘I would have gone there yesterday.’ b. pooy-ITTUND-aayir-unn-illa. 20-PERF-be.PAST-unn-NEG ‘(He) had not gone.’ c. paranfi—ukoNDinkk-unnu. say-PROG-PRES *(He) is saying.’ d. paathannal pathippikkape{{ukontirunni{tuytaakanam. (see Asher and Kumari 1997, 304) lesson-PL learn-CAUS-PASS-PROG-PERF,-DEB-PRES ‘Lessons must have been being taught.’ AUSTRIC

Kammu (Austro-Asiatic): Moodgpecch act MOOdinreatis / ASPhabituat ASPperfece V (See

Svantesson 1994, 271ff) (19) a. 0cd=9 praa pd mah kaal kmra rdot taa kaan. Isg IRR NegPERF eat food before wife come to home ‘T will not have eaten before [my] wife comes home.’

OVERT FUNCTIONAL HEADS IN INDIVIDUAL LANGUAGES 159

b. 6 ki pd mah héoc yém kmra rot taa kaan.® isg HAB eat food PERF when wife come to home ‘T have usually eaten when [my] wife comes home.’ c. béc mée héoc pip yon 0? Q you PERF meet father 1sg ‘Have you ever met my father?’ Thai (Austro-Ta1): Modgpistemic T(Future) Aspprogressive ASPprospective Voice V

(20) a. khaw kamlang thutk nansv¥phim wicaan. (Wongbiasaj 1979, 211) he PROG PASS newspaper criticize ‘He is being criticized by the newspaper.’ b. kh4w kamlang ca? pay miawaannii. (Steele 1975, 46) he PROG PROSP go yesterday ‘He was going to go yesterday.’ c. khong ca? kamlang thamngaan naj héng samt. (Pornsiri Singhapreecha, personal communication) EPISTEM FUT PROG do.work in library ‘T will probably be working in the library.’ d. lom khong ca? kamlang phat.’ (Steele 1975, 46) wind EPISTEM FUT PROG blow ‘The wind must be blowing.’ e. khaw kamlang ca? thuuk wicaan. (Pornsiri Singhapreecha, personal communication) he PROG PROSP PASS criticize ‘He is about to be criticized.’

AUSTRONESIAN

Malay (Sundic): T(Anterior) Asppertect / ASPprogressive ¥

(21) a. bila gua sampai dia sudah habis pergi. (Baxter 1988, 123) when l|sg arrive 3sg ANT PERF go “When I arrived he had already gone.’ b. dia sedang lihat buku. (Baxter 1988, 130) 3sg PROG read book ‘He is reading a book.’ Kwaio (Eastern Malayo-Polynesian): TFuture) Asppertect / ASPproximative V (see Keesing

1985, 118ff) (22) a. ta-goru bi’i aga-si-a.!° FUT-AGRs PROXIM see-TR-AGRo “We'll see it soon.’ b. ngai e leka no’o. he AGRs go PERF ‘He has gone.’ Ponapean (Micronesian): T(Future) Aspyapituat / ASPpertect ASPprogressive W (See Rehg

1981, 269ff)

160 APPENDIX 2

(23) a. i pahn kin kangkang rais.!! I FUT HAB eatPROG rice ‘T will (habitually) be eating rice.’ b. Seriet lingilingeringer-ehr. that child angryPROG-PERF ‘That child has been being angry.’

Kiribatese (Micronesian): T(Future) Aspyerminative “SPiterative V (See Groves, Groves,

and Jacobs 1985, 78) (24) Ena tian naanako Tarawa. He FUT TERMIN go-ITER to T ‘He will be finished going to T.’

Anejom (Melanesian): T(Past) T(Future) Asppertece V (see Lynch 1982, 119ff) (25) a. Is mian hag afiak. PAST PERF eat I ‘T have eaten.’

b. Is ika aen is pu apam imraii. PAST say he PAST FUT come tomorrow ‘He said he would come tomorrow.’ Samoan (Polynesian): Moodgpeech act P(Future) Mood ,reatis ASPperfect ASPprogressive V (S€€

Mosel and Hovdhaugen 1992, 354, 360ff) (26) a. “Ae a lava pea sei e alu atu e ai lelei.!* but EMPH EMPH Q FUT IRR 2sg go DIR GEN eat good ‘But if you go and eat well...’ b. Nofo mai i lalo sei o ta talanoa. Stay DIR LD down IRR PROG lincIDUAL talk ‘Sit down here so we can have a talk.’ c. ‘A ‘o lae ‘olo’oua tagi le tama. but PRES this PROG-PERE cry art youth ‘The youth was weeping .. .’ Tokelau (Polynesian): T(Anterior) Asppertect ASPprospective V (see Vonen 1994, 385ff).!°

(27) a. ko au kua k@ fano. PREP Isg PERF PROSP go ‘I have become about to go.’ b. Na kua pa koe ki Niu Hila? ANT PERF reach you PREP New Zealand ‘Have you ever been to New Zealand?’

Big Nambas (Oceanic languages): T(Future) / Mood gp reais Mod T(Anterior) Asp proxumative ASPyepetitrve Vv (see Fox 1979, 64)

(28) a. p’e-pah-ma. FUT-PROXIM-come ‘T shall soon come.’

OVERT FUNCTIONAL HEADS IN INDIVIDUAL LANGUAGES 161

b. 1-k-d-a-ruh. he/realis-NECESS-ANT-NEG-run away ‘He mustn’t run away yet.’ c. a-d-a-v-mur-ma. They/realis- ANT-NEG-pl.-REPET-come ‘They haven’t yet ever come.’ AUSTRALIAN

Walmadj ari: T(Past) Mood, realis ASD continuative V

(29) natjita pa yan-an-ta-la. (from Hudson 1976, 656) not he go-CONT-IRR-PAST ‘He didn’t keep on going.’ N § Ly ambaa: Moodaeyajuatve T(Future) Mod ASP durative ASP progressive / ASPrepetitive V (see

Donaldson 1980, 183ff, the example below, and chapter 4) (30) yana-buna-biya-y-aga. go-back-NECESS-CM-FUT ‘We will have to go back’

INDO-PACIFIC (PAPUAN LANGUAGES)

Fore: Mood ecch act F(Past) ASPierminative V (See Foley 1986, 145)

(31) na-kai-?ta-1-e. eat- TERMIN-PAST-3SG-DECL ‘He had finished eating.’ M enya. Mood neech act T(Present) ASP progressive Vv

(32) yakaé hinué 4-w-g-n-atig-n—qaqu-i, ... (see Whitehead 1991, 266) bridge eye ?-AGR-strike-DETRANS-PROG-PRES-AGR-DECL “While we were looking at the bridge, .. .’ Sanio-Hiowe: Mood peech act 1(Past)/T(Future) Aspperfect V (See (S9a—b) of chapter 3) Tauy a. Mood secch act Modenstemic T(Future) ASP habitual ASP perfect! ASP durative ASP pearly /

ASP progressive! ASPconative ¥ (See §3.1 above and MacDonald 1990, §3.3.2.1, and n. 64

of chapter 4) (33) a. ya ni-we-pope-e-?a. Isg eat--CONAT-HAB-1/2—DECL ‘T always try to eat.’

b. sawi ni-?afe-a-?a. banana eat-PROG-3s¢g-DECL “She is eating bananas.’ Cc. mene-tei-pe-i-na. stay-DUR-HAB-3pl-INCONS

‘They stayed for a long time and...’

162 APPENDIX 2 Una: Modenistemic 1 Mod T(Antertor) (see (9a-b) of chapter 3, and (34) from Louwerse 1988, 66)

(34) E-an-se. say-ANT-PAST. lsg ‘I had said.’ Yareba: T AsPpabituat ASPfrequentative W (see Weimer 1972, 61)

(35) yau-r-edib-eb-a-su. sit-class marker-FREQ-HAB-PRES-3sg/masc ‘He (habitually and repeatedly) sits down.’ Wahgt: Moodgpeech act 1(Future) Asppapituay V (see Phillips 1976, 97)

(36) na-pi-!-tang-n-a-mbii-mo? NEG-hear-CM-HAB-CM-FUT-DUAL-Q “Will the two of them always not hear?’ NA-DENE

Navajo: T Asp V (Speas 1991a,b) AMERIND!4

Canela-Crahé (Cariban): Mood, jccech act T(Past) ASPhabituat ASPprogressive V (See Popjes

and Popjes 1986, 157, 182)

(37) a. xacapi te po curan? Q C. PAST deer kill ‘Did Capi kill a deer?’ b. pé wa ajco apu to hane.!° PAST(Distant) Isg HAB PROG do thus ‘I always used to do that.’ Dieguenio (Hokan): Mood speech act MOdepistemic P(Future) V

(38) a. wa:m-x-kox. (Bybee 1985, 179) he.goes.away-FUT-must.be ‘He might go.’ b. wa:m-x- kox-a? (Bybee 1985, 174) go-FUT-PROBABLE-Q ‘Might he go?’ Hidatsa (Siouan): Moodgyidential T(Past) / ASP frequentative ASPinchoative V

(39) a. wirai 4paari ki stao wareac. (De Groot 1995, 40) tree 3sg grow INCH (remote)PAST EVID ‘They say the tree began to grow a long time ago.’ b. wio i hirawe ki ksa c. (Hengeveld, forthcoming b, ex. 42) woman 3sg sleep INCH FREQ EVID “The woman fell asleep again and again.’

OVERT FUNCTIONAL HEADS IN INDIVIDUAL LANGUAGES 163

Ika (Chibchan): Modepistemic T(Future) Aspinceptive ASPprogressive VW (see Frank 1990, 108,

57) (40) a. Kant mus-An nA—ngua ni. cane grind-PROG 1s-FUT CERTAINTY ‘I will grind sugar cane.’ b. Ingt-ri tSout8o kanak-am-pana keiwi. little-TOP afraid become-PROG-INCEPT right away “He began to get scared.’ Macushi (Cariban): T(Past) Aspgnany (Anterior)/Aspperfec/ ASPfrequentative ¥ (see Abbott

1991, 113ff) (41) a. to’ erapami-“pi wani-‘pi. 3pl arrive-ANT be-PAST “They had arrived (before).’ b. atti-sa’ wani-‘pi. 3sg.g0-PERF be-PAST ‘He had gone.’ (He was gone.)'® c. aa-ko’man-pt’-sa’. 3sg-remain-FREQ-PERF ‘He has remained (repeatedly).’ d. i’-po’-piti-tu’ka-‘pi-i-ya. 3-whip-ITER-finally-PAST-3-ERG ‘He finally whipped him.’ (mbabura) Quechua: T(Past) Asp perfect/ ASPaurative ASPprogressive ASPprospective Causative

V (Cole 1985, 145, 152)!” (42) a. riku-shka-rka-ni. see-PERF-PAST-1 ‘I had seen.’ b. shamu-ju-shKa-ni. come-PROG-PERF-1 ‘I have been coming.’

Ute (Uto-Aztecan): Mood, yigentiat T(Past) ASPpabiar V (see §3.1 and Givén 1982a) Yr avapai (Hokan): ASP progressive ASP repetitive Vv

(43) @ala-c mi:-yi-km. (see Kendall 1976, 30) 9. cry-REPET-PROG ‘@. is crying again.’ Waorani (South American): Mood, jecch act MOdepistemic / T(Past) ASpinceptive ¥ (See Peeke

1994, 276, 288)

(44) a. ...tdbégd wae-kae-k4-i-pa. (s)he cry-INCEPT-3sg-EPISTEM-DECL ‘(S)he will surely be sorry.’ b. Wii, bot6 kae-dabai-i—ta-bo-pa. not, | eat-NEG be-PAST-Isg-DECL "No, I didn't eat.’

164 APPENDIX 2 CREOLES

Berbice Dutch Creole: T(Past) T(Future) Asppabiwar T(Anterior) Aspyepeutve V

(see Kouwenberg and Robertson 1988; Robertson 1990; Kouwenberg 1994, chap. 3)!®

(45) a. ek wasa ku-te en or twe fan eni an tem enl. Isg PAST FUTGrr) catch-ANT one or two of 3pl and tame 3pl ‘I would have caught one or two of them, and tamed them.’ b. Alwes, di hari das jenda, baja di toro. always, the hair HAB be = there cover-IMPF the face ‘The hair always covers the face.’

Guyanese Creole: Mod, pi stemic T(Past) T(Future) Aspyapina/T (Anterior) ASPaurative ASPprogressive V (See $3.3 and Gibson 1986, 1992) Haitian Creole: T(Past) T(Futur e) ASPhabitual ASP retrospective ASPprogressivel ASP prospective Vv

(see §3.3 and references cited there) Kristang: T(Anterior) ASPpertect / ASPprogressive V (Baxter 1988, chap. 5)

(46) a. Kora yo ja chega eli ja ta kumi. when Isg ANT arrive 3sg ANT PROG eat ‘When I arrived he was already eating.’ b. Kora yo ja chega eli ja kaba bai. when Isg ANT arrive 3sg ANT PERF go “When I arrived he had already left.’ Louisiana Creole: T Mood reais MOd ASP perfect ASPprogressive © (Rottet 1992, 274 and

n. 12) Ndyuka: T(Past) T(Future) Moodireans Mod Asppyogressive V (See Huttar and Huttar 1994,

319)

(47) Ibe o sa poi (fu) nyan ete 2sg PAST FUT IRR can (for) eat yet “Would you have been able to eat already?’

Nigerian Pidgin: T(Past) T(Future) T(Anterior) Aspprogressive YW (see Ofuani 1981, 1982)!° Seychelles Creole: T(Past) T(Future) T(Anterior) AsPpyetrospective ASPprogressive V (Papen

1978, 362; see also Corne 1977, 94ff, and S. Michaelis 1993, LO3ff) (48) Za tian fek pe maze. J. PAST FUT ANT RETRO PROG eat ‘J. would just have been eating.’

Sierra Leone Krio: T(Past) T(Future) T(Anterior) Aspprogressive V Jones 1990, 849)

(49) Bin go don de it. Isg-PAST FUT ANT PROG eat ‘T should have been eating.’

OVERT FUNCTIONAL HEADS IN INDIVIDUAL LANGUAGES 165

Sranan, T(Past) T(Future) Mod T(Anterior) Aspprogressive V (See $3.3, Seuren 1981, 1983; and Adamson and Smith 1995, 229) LANGUAGE ISOLATES

Basque: Moodgyidentiaa TY MOd ASPpapimat ASPpertece VOICE V (see (S8b) of chapter 3, and

the examples and references cited in n. 20 and n. 59 of the same chapter; and the following example): (50) eda-n ohi du. (Ortiz de Urbina 1989, 141) drink-PERF HAB AUX “He usually drinks.’

Coahuilteco: Mood speech act F(Past) / T(Future) Mood, reais MOyoutionat ASPrepetitive ASPpretensive V (See Troike 1996, 658)?°

Zuni:*' T(Past) T(Future) Asp V (see Nichols 1993, 99, 104) progressive

(51) a. Pa:n-uwa-nkya. go-FUT-PAST ‘He should have gone.’ b. Hom-?an ten-e:-?a. {sg.ACC-DIR sing-PROG-PRES ‘He is singing for me.’

This page intentionally left blank

NOTES

Chapter 1 For their judgments and comments on the material in this chapter, I wish to thank Adriana Belletti, Paola Beninca’, Gerhard Brugger, Anna Cardinaletti, Carmel Coonan, Peter Culicover, Rose-Marie Déchaine, Giuliana Giusti, Hubert Haider, Marie Christine Jamet, Richard Kayne, Victor Manfredi, Rita Manzini, Friederike Moltmann, @ystein Nilsen, Cecilia Poletto, JeanYves Pollock, Luigi Rizzi, Michal Starke, Marie-Therése Vinet, and Raffaella Zanuttini. 1. For suggestions about analyzing (some) adverbs along these lines, see Sportiche (1993, 1995b).

2. If the verb can cross over an adverb within a larger constituent, as Sportiche assumes (and I assume in §1.4 for certain special cases), this may actually be rather delicate to establish. 3. An additional argument for treating adverbs as heads of XPs in a specifier position— rather than as heads taking an extended projection of V as a complement—may come from the “grammar” of code switching. As Mahootian and Santorini (1996, 472ff and n. 9) observe, a head always dictates the position of the complement (irrespective of the “code” of the complement). But with adnominal and adsentential adjuncts (adjectives and adverbs), anything goes, and this is another reason not to see what follows the adverb as its complement. 4, In standard Italian, this element has no autonomous negative force in postverbal position and must co-occur with the preverbal negative element, non (*Vengo mica ‘I come not’; see Non vengo mica). See Cinque (1976) for some discussion of its pragmatic import, which, to judge from Espinal (1993), seems to correspond rather closely to that of Catalan pas and to what Zanuttini (1997) calls “presuppositonal” negation (see chapter 5, §5.4). On mica preceding a finite verb, see chapter 2 (§2.2). On the position of guére in French, see n. 19.

167

168 NOTES TO PAGES 5-6

5. See also the order of the two adverbs when they constitute an elliptical answer: Tu fumes? Généralement pas (*Pas généralement). Irrelevant, from this point of view, is the well formedness of // ne fume pas, généralement (where généralement is ““de-accented”’ and takes scope over pas; see §1.4 and $1.6 for relevant discussion). Généralement and other habitual adverbs in French (and in Italian) have two additional usages that should be kept distinct. One is a “focusing” usage (for which see note 12 and $1.6), as in C’est généralement a Marie qu'il accorde toutes ses attentions ‘It is generally to Marie that he dedicates all his attentions.’ The other, observed in Mgrdrup (1976, 321), is a usage in absolute initial position as a “domain” adverb (for which see note 41).

6. Although the past participle can precede both mica and gid separately (Non ha mangiato mica “He hasn’t eaten not’; Ha mangiato gia ‘He has eaten already’), it cannot apparently precede the two together (“Non ha mangiato mica gid) unless some other constituent follows (see Non ha mangiato mica gia tutto “He hasn’t eaten not yet everything’). This may be because gia cannot be the only element falling under the scope of negation (see *Non ha mangiato gia “He hasn’t eaten already’). The reinforced variant, di gia, which can bear focal stress, can, however: Non ha mangiato (mica) di gid. 7. Unlike Italian, in which the order gia mica is never possible, the order déja pas as such is attested, but trrelevantly for our concerns, as déja here appears to form a constituent with (to be in the specifier of) pas. Togeby (1984, 219) notes: “Déja peut renforcer la négation,” citing the following example: Ce n’est déja pas si mal. ‘It 1s already not so bad.’ Quite gen-

erally, pas, in French—as opposed to mica in Italian—may take specifiers (Elle ne proteste pas a haute voix, ou presque pas. ‘She does not protest loudly, or almost not’ [Togeby 1984,

259; Déprez 1995] versus Italian *..., o guasi mica. Thus, the possibility of [déja [pas]], [toujours [ pas]] (see note 14), and so forth, in French, contrasts significantly with the impossibility of [gia [mica]], [sempre |mica]}], and so forth, in Etalian. 8. There is evidence that mica pi can also form a single constituent, with the negative adverb mica directly modifying (in the specifier of) piu. For example, the sequence can be fronted under Focus Movement (MICA PIU l’ho visto ‘Not any longer have I seen him’) and can constitute a well-formed answer by itself (Lo vede ancora? Mica piu. “Does he still see him? Not any longer’). This is not the case with other combinations of adverbs, such as mica mai, ‘not never’ (*MICA MAI I’ho visto ‘Not never have I seen him’ Lo vede ancora? *Mica mai ‘Does he still see him? Not never’), mica ancora, ‘not yet’ (*MICA ANCORA Vho ricevuto ‘Not yet have I received it? Lo hai ricevuto? *Mica ancora ‘Have you received it? Not yet’). These systematic contrasts follow if mica pia—but neither mica mai nor mica ancora— may function as a single constituent. This, of course, does not necessarily imply that mica piu always is a constituent. The fact that the past participle can intervene between the two elements indicates that it need not: Non l’ha mica fatto piu, da allora ‘He hasn’t not done it any longer, since then.’ On mai piu necessarily forming a single constituent, see n. 16. 9. The point is not really affected by the existence of few cases of an infinitival verb preceding pas in literary works (see Engver 1972, 18ff). 10. Except in the literary variety referred to in the previous footnote. The fact that pas and plus in French cannot co-occur (as opposed to mica and piu in Italian) appears to be related to the general ban against the co-occurrence of pas with inherently negative elements U1 n’a (*pas) rien dit “He said nothing; // n’a (*pas) vu personne ‘He saw nobody’, etc., versus Italian Non ha mica detto niente; Non ha mica visto nessuno; see Acquaviva 1995 and references cited there). As pointed out to me by Anna Cardinaletti, plus in French (as opposed to pi in Italian) appears to carry an autonomous negative force in that it may constitute the sole negation in such small clause contexts as the following: Je (*ne) considére |Jean plus capable de faire ca] ‘I consider J. no longer capable of doing that’ versus *Considero J. piu capace di far

NOTES TO PAGES 6-7 169 questo. Consistent with this is the possible use of plus without ne mentioned in Pollock 1989, 418, n. 44 (C'est bien de plus fumer ‘It’s good to no longer smoke’ versus E’ una buona cosa *(non) pia fumare, in Italian) and the use, noted in Muller 1991, 291, of plus as an autonomous negative answer (Ju revois Marie? ?Plus “Will you see M. again? No more’), which contrasts again with Italian pit (Rivedi Maria? *(Non) pit). In Québec French, pas can cooccur with rien and personne but, for some reason, still cannot co-occur with plus (MarieTherése Vinet, personal communication). This could be a problem. This contrasts with the

possibility of the sequence pa pi in Piedmontese and other dialects of northern Italy (see Zanuttini, 1997, chap. 3). For an interesting analysis of negation in different French varieties and French-based Creoles, see Déprez 1995. 11. Though heavy, the co-occurrence of the four adverbs in Italian appears to be possible:

(i) A quest’ora, non ha solitamente mica gia pit fame. ‘At this time, he isn’t usually not already any longer hungry.’ 12. The apparently inconsistent order of Hanno sempre gia mangiato ‘They have always already eaten’, with sempre’s unexpectedly preceding gid, can be reconciled with (12) if gia is also able to modify the participle (and whatever follows it), much as other “focusing” adverbs (Rooth 1985, Bayer 1996) like solo ‘only’ and anche/perfino ‘even’ (for their structural location, see $1.6). This seems supported by the fact that gid can be fronted together with the participle plus whatever follows it (Gia mangiato le paste, credo che non abbiano ‘Already eaten the cakes, I believe they have not’) while “nonfocusing” adverbs cannot (*Spesso mangiato le paste, credo che non abbiano ‘Often eaten the cakes, I believe they have not’; *Mai parlato a GIANNI, credo che abbia ‘Never spoken to G. (focus), I believe he has’). The “focusing” usage of already is also apparent in the following Norwegian sentences provided by @ystein Nilsen, where, for V/2 reasons, allerede must form a constituent with the following DP and the AdvP, respectively:

(i) a. Allerede Jon gjennomskuet dette komplottet. ‘Already J. figured out this plot.’ b. Allerede i gar gjennomskuet Jon dette komplottet. ‘Already yesterday figured out J. this plot.’ See §1.6 for similar “focusing” uses of “higher” adverbs like probabilmente ‘probably’, fortunatamente ‘luckily’, and so forth. Another possibility, which will be discussed in more detail in chapter 4, is that sempre ‘always’ and other “quantificational” adverbs (for example, abitualmente ‘usually’, di nuovo ‘again’, and frequentemente ‘frequently’), are actually generable in two distinct, quantificational, “zones” (quantifying over different domains, corresponding to the event and the process, respectively). 13. The conjecture in the previous note about the possibility for gid, in its “focusing” usage, to also directly modify a lexical (or functional) constituent may extend to the apparently inconsistent order toujours déja of such examples as Quand je le rencontre, il est toujours déja ivre ‘When I meet him, he’s always already drunk.’ 14. Togeby (1984, 221) observes that the opposite order is also attested (with the meaning of “encore’): On ne démarrait toujours pas “We were not beginning yet’. We must assume that toujours, here, is in the Spec of pas (see n. 7, above). Confirmation of that comes from the fact that pas can be separated from toujours when it precedes it, as in (ia), but not when it follows it, as in (ib):

Gi) a. Ne pas étre toujours invité est normal. ‘Not to be always invited 1s normal.’ b. *Ne toujours étre pas invité est normal. ‘Not to be yet invited is normal.’

170 NOTES TO PAGES 7-8 15. For arguments that tutto occupies a derived position in (23) as well, see Belletti (1990, 77ff£) and references cited there. See also Cinque (1992a). 16. Pia (but neither mica nor gia) can also occur after a verb complement when modified, coordinated, or focused (the latter, in fact, only marginally):

(i) a. Da allora, non diede il suo aiuto a Carlo mai PIU. ‘Since then, he gave his help to C. never any longer.’ b. Da allora, non da il suo aiuto pit o quasi piu. ‘Since then, he gives his help no longer or almost no longer.’ c. ??Da allora, non da il suo aiuto a Carlo PIU. ‘Since then, he gives his help to C. no longer (focus).’ As a consequence of that, it may be found to follow completamente ‘completely’, tutto ‘everything’, and bene ‘well’, differently from the unmarked (unfocused) pia examined earlier (for further discussion on this apparent VP-final position, see $1.4):

(i) a. Non perdera completamente la testa pit. o quasi pit. ‘He will not completely lose his mind any longer or almost any longer.’ b. Non perdera tutto mai PIU. ‘He will not lose everything never any longer.’ c. Da allora, non si comportd bene mai PIU. ‘Since then, he did not behave well any longer.’ Mai pitt can only occur as a single constituent (with mai in the specifier position of piiz). See the ungrammaticality of (411) noted in Belletti (1990, 59), with mai separated from piu by the past participle, and the discussion in the text of jamais plus and plus jamais in French:

(iii) *Maria non ha mai parlato pit con Gianni. *M. hasn’t ever spoken any longer with G.’ 17. This sentence is (irrelevantly) grammatical in the reading in which parzialmente modifies just bene, plausibly in its specifier position (“partially well/almost well”). Indirect confirmation that completamente (parzialmente) precedes tutto derives from the fact that the past participle cannot be to the right of tutto (*Lui ha tutto perso “He has everything lost’) though it can be to the right of completamente/parzialmente, as well as to its left (see Lui ha completamente/parzialmente perso tutto ‘He has completely/partially lost everything’; Lui ha perso completamente/parzialmente tutto). This becomes understandable if completamente/ parzialmente is higher than tutto and if the past participle has to move to a head to the left of tutto, but may fail to move any higher (see chapter 2). 18. I thank Marie-Therése Vinet for these judgments, As Marie Christine Jamet pointed out to me (personal communication), with other lexical choices, both the order complétement tout and the order tout complétement appear to be possible:

(i) a. Tu as complétement tout refait? b. Tu as tout compléetement refait?

They differ, however, in interpretation. In (1a), complétement appears to modify both the verb and the object (in which case, it can be naturally answered with Non! Je n'ai fait que la cuisine). In (ib), the adverb appears to modify just the verb (it can be naturally answered with Non! Je n’ai fait que la peinture). This may suggest that in (ib) complétement is reinterpreted as a “focusing” adverb, which modifies just the verb. 19. Guére does not appear to occupy the same position as pas. It follows plus (Togeby 1984, 269; Muller 1991, 271 ff), and toujours, while preceding complétement and tout (also see Pollock 1989, 413ff).

NOTES TO PAGES 8-9 171

(i) Je ne pourrai (*guére) plus (guére) venir. ‘ET could not come any more (much).’

Gi) Il n’a (*guére) toujours (guére) accepté. ‘He has not always accepted (much).’ (iit) Il n’a (?guére) completement (*guére) perdu la téte. ‘He has not (much) completely lost his mind.’

(iv) Il n’a (2guére) tout (*guére) mangé. “He has not (much) eaten everything.’ { thank Marie Christine Jamet and Jean-Yves Pollock for sharing their intuitions with me on this point. Relevant here is also the fact that a lexical infinitive can precede guére (Togeby 1984, 264), but not pas. Guére wouid seem to be the negative counterpart of quantity adverbs like beaucoup, peu, trop, and so on, but, as noted in Vikner (1978, 106) it does not occupy the same position. While the latter are found between the passive and the lexical past participle (Ce livre a été beaucoup/peu lu l’année derniére “This book has been much/little read last year’), guére has to precede both (Ce livre n’a guére été lu l'année derniére “This book has not been read last year’). All this suggests the presence of a second NegP, lower than that hosting pas/mica, apparently to be identified with that hosting no (as opposed to minga/mia) in Milanese and Pavese (see Zanuttini, 1997, chap. 3), and possibly nicht in German (Brugger and Poletto 1993). On the position of Norwegian ikke (apparently the same as Italian mica), see §1.7. For the postulation of different NegPs in different positions of the clause, see the discussion in chapter 5 and m Zanuttini (1997, chap. 3). 20. Onaffatto, see Acquaviva (1995). Onno, Poletto (1998). Belletti (1990, 39ff; 1994b) interprets “affirmative” ben and pur (see “affirmative’ bien in French) as the positive counterpart of negative adverbs. Indeed, they appear to follow habitual adverbs and to precede gid, exactly like mica, affatto, and so on.

(i) a. Lui ha di solito ben/pur mostrato di apprezzare la musica. “He has usually indeed shown he appreciates music.’ b. *Lui ha ben/pur di solito mostrato di apprezzare la musica. ‘He has indeed usually shown he appreciates music.’

Gi) a. Avrai ben/pur gia cenato. “You will have indeed already eaten.’ b. *Avrat gia ben/pur cenato. ‘You will have already indeed eaten.’ See, however, chapter 5 (§5.4) for the possibility that affirmative adverbs occupy a position contiguous to, but distinct from, negative adverbs. Another affirmative adverb, which corresponds to the affirmative use of do in English (Pecoraro-Pisacane 1984, 62), is si ‘yes’, which, besides the position of ben/pur, can occupy other positions, as shown in Poletto (1998). 21. This positioning of poi is observed in a number of northern Italian dialects in Zanuttini (1997), 22. The close relationship between the two terms is further shown by the fact that their English equivalents, already and yet, form in questions a suppletive pair of the some/any kind (Traugott and Waterhouse 1969), in which “already expects an affirmative answer whereas yet leaves open whether the answer is negative or positive” (Quirk et al. 1985, 581):

(i) a. Have you already seen him? b. Have you seen him yet? 23. In French, encore precedes plus, as in Italian, but can apparently either follow or precede pas, though Togeby (1984, 219) says that “Vordre pas encore est bien plus courant que encore pas’:

172 NOTES TO PAGE 9

G) a. Il n’a encore plus rien recu. ‘He has not yet any longer received anything.’ b. *II n’a plus encore rien recu. ‘He has no longer yet received anything.’

Gi) a. Elle ne l’a pas encore lu. ‘She it has not yet read.’ b. Elle ne l’a encore pas iu. ‘She it has yet not read.’ As usual, the unexpected order (encore pas) should necessarily have encore in the Spec of pas (a possibility open to pas in French, though not to mica in Italian, as noted in n. 7). The test of inserting a verb in between the two goes in the expected direction (see (iiia—b)),

although the judgment appears to be less clear here than with déja pas / pas déja, toujours pas / pas toujours: (111) a. Ne pas avoir encore lu ce livre est inadmissible. ‘Not to have yet read this book is inadmissible.’ b. ??N’encore avoir pas lu ce livre est inadmissible. On ancora/encore preceding the (negated) finite verb, see chapter 2 (§2.2). 24. Ancora signals the “continuation” of a certain state or process while pii indicates its “termination” (see chapter 4 for their relation to Continuative and Terminative aspect, respectively). This (“positive”) ancora (= still) has a different interpretation from the ancora of non... ancora (= yet), entering, as noted, into a different suppletive set. This is not to say that the two uses, and the use in which ancora means ‘again’ (see n. 26), are not (semantically) related (see Vegnaduzzo 1997). For the intricate interrelationships among these adverbs and gia and pitt in a number of languages, see van der Auwera (1991, 1993, forthcoming), and references cited there. 25. French, once again, allows both orders (see (ia—b)), but the impossibility of inserting anything between toujours and encore (see (iia—b) provides evidence that, in the unexpected order (toujours encore), toujours is in the Spec of encore:

(i) a. Jl aencore toujours les meilleurs chances. ‘He still always has the best chances.’ b. Il a toujours encore les meilleurs chances. ‘He always still has the best chances.’

(ii) a. Encore avoir toujours les meilleurs chances. b. *Toujours avoir encore les meilleurs chances, 26. Differently from pia, ancora cannot co-occur with gia, possibly for semantic reasons (the same is true of déja and encore, in French). It can, however, co-occur withpoi ‘after’, which we have seen belongs to the same class as gid, and it follows it, as expected:

(i) a. CeVhai poi ancora, quel libro? ‘Do you after still have it, that book?’ b. *Ce hai ancora poi, quel libro? ‘Do you still after have it, that book.’ The order sempre > ancora is also possible, but with ancora meaning ‘again’ rather than ‘still’. Note that in this case, though not in the one where ancora means ‘still’, ancora preferably follows the past participle:

(i) a. ??Lui lo ha sempre ancora rivisto. ‘He has always seen him again.’ b. Lui lo ha sempre rivisto ancora. ‘He has always seen him again.’ See chapter 4 for some discussion of ancora = ‘again’, and the corresponding Repetitive aspect found in many languages, which indeed occupy a low position. 27. The sequence piu mai was possible in Old Italian, along with mai piu (see Battaglia 1986, 485), and is found in literary Jtalian up to the nineteenth century, as Paola Beninca’

NOTES TO PAGES 9-11 173

reminded me. (See U. Foscolo’s poem A Zacinto: “Né piri mai tocchero le sacre sponde.” *No longer ever will I touch the sacred banks.’) Interestingly, in Old Italian, but not in Modern Italian, the sequence mai piu can appear separated by a finite verb (see Battaglia 1986, 485). A reasonable conjecture is that this possibility is crucially related to the availability of the sequence pit mai, in which the two do not necessarily form a constituent, plus fronting (focalization) of the sole mai. (If no independent Focalization of the specifier of an AdvP is possible, mai cannot instead be separated from piu in Modern Italian.) 28. See, for example, Wartburg and Zumthor (1958, 63): “usage moderne préfére Vordre plus jamais” (cited in Engver 1972, 25). 29. There is evidence that plus jamais can also form a constituent, with plus in the specifier position of jamais. The sequence can appear preposed (focalized) in front of the subject, not differently from jamais plus (see Ga—b), from Muller 1991, 276), and can constitute an isolated answer (see (ii), again from Muller 1991, 339):

Gj) a. Plus jamais il ne m’a écrit. ‘No longer ever he wrote to me’. b. Jamais plus il ne m’a écrit. ‘Never any longer he wrote to me.’ (1) Tu y retournes? Plus jamais! (in addition to Jamais plus!) You go back there? No longer ever! Also, the fact that the infinitive may precede plus jamais (Engver 1972, 24) can be taken as additional evidence for the location of plus in the Spec of jamais. This is because an infinitive can precede jamais, but not plus (for most speakers). Given this, the impossibility of the sequence pitt mat in Italian is all the more surprising in that it must now be excluded both as a constituent and as a nonconstituent. I have no suggestions to offer. 30. Pointing to the same conclusion is the contrast, observed by Anna Cardinaletti, between (1a) and (ib), which shows that niente has to precede the (inverted) subject position (which is presumably in Spec VP):

(i) a. Non ha fatto niente nessuno. ‘Not has done nothing nobody.’ b. *Non ha fatto nessuno niente. ‘Not has done nobody nothing.’ Niente also contrasts with ordinary DPs, as can be seen by comparing (1a) with (a1):

(ii) %Non ha fatto nessuna/questa cosa nessuno. ‘Not has done no/that thing nobody.’ While (11) is (marginally) possible only with nessuno in focus (and the rest presupposed), (ia) is also possible in a context where nothing is presupposed. 31. But this might not be correct. In the rare contexts in which molto/poco combine with bene (without directly modifying it), they seem to precede bene: (?)Ballava molto assai bene anche il fratello ‘Even his brother danced much very well’ versus *Ballava assai bene molto anche il fratello. On the apparent possibility of coordinating molto and bene (Ballava molto e bene ‘He danced much and well’), see the discussion in chapter 4 (§4.25, n. 72). Vinet (1995a) reports a sentence where beaucoup is to the left, rather than to the right, of tout. See (4) (= her

ex. (20a), which, however, is not accepted by all speakers:

Gi) J’ai pas toujours beaucoup tout bien nettoyé. ‘T have not always a lot everything well cleaned.’ Perhaps, for certain speakers, beaucoup can also belong to the class of complétement, which we saw is to the left of tout. 32. On the two positions occupied by spesso (and related adverbs), neither of which corresponds to that hosting sempre (and mai), see §1.4 and chapter 4.

174 NOTES TO PAGES 11-12

It is interesting to note that the semantic contribution of mica ‘not’, non... ancora ‘not ... yet’, pia ‘any longer’, and ancora ‘still’ is not truth-functional but presuppositional (see Cingue 1976 and Zanuttini, 1997, chap. 3, on mica, and Ladusaw 1978 on yet, anymore, and still). The same holds of gid ‘already’: Gianni é gid arrivato ‘G. has already arrived’ asserts the same thing as Gianni é arrivato ‘G, has arrived’ (without gid), and merely adds the presupposition (or conventional implicature) that “G.’s arrival was prior to now (and perhaps earlier than expected).” Ladusaw (1978) assigns again to the same class. Interestingly, ancora, qua ‘again’, in Italian, will be seen in chapter 4 to fall in the same “presuppositional” “space.” 33. Bellert’s “domain adverbs” correspond to Kuno’s (1975) and Ramat and Ricca’s (1994, 313) “adverbs of setting,” to Schlyter’s (1977, 81ff) “adverbes cadre” (which comprise “adverbes de point de vue,” “adverbes de temps,” et “adverbes de norme”’), to Quirk et al.’s (1985) “viewpoint adverbs,” and to Parsons’ (1990, 211ff) “frame adverbials.” For further discussion, see n. 41 and T. B. Ernst (1984, 39-~56). 34. For the distinction between “evaluative” and “modal” (the latter called “epistemic” by Huang 1975, 46ff; T. B. Ernst 1984, 56-64; Travis 1988, 299; Baker 1989, 279; Nuyts 1993), see also Schreiber (1971) and Parsons (1990, 62ff). 35. By transitivity, “subject-oriented” adverbs should follow “modal” adverbs; which is what we find (as already noted above with reference to Jackendoff 1972, 89). 36. Forse in (46a) can (irrelevantly) also be construed as directly modifying the “subject-oriented” adverb. Care should also be taken not to give forse, in (46b) (or, for that matter, per fortuna in (48b), francamente in (52b), etc.) a “parenthetical” intonation, which appears to partially rescue the sentence. On this parenthetical usage of AdvPs, see §1.6. 37. “Evidential” adverbs (Chafe 1986; Fraser 1996) such as allegedly, obviously, clearly, evidently, etc.), which are sometimes assigned to the class of “modal” (epistemic) adverbs, should perhaps be assigned to a distinct class. This is suggested by the fact that, whereas the relative order probabilmente > evidentemente is unacceptable ((ia)), the opposite order appears possible (I differ here from Belletti 1990, 130, n. 29):

(i) a. *Probabilmente Gianni ha evidentemente deciso di partire. ‘Probably G. has evidently decided to leave.’ b. (?)Evidentemente Gianni ha probabilmente deciso di partire.

Bowers (1993, n. 13 and p. 607) mentions such sentences as John obviously will probably soon learn French and Clearly John probably will quickly learn French perfectly, noting for the latter that no other order of the adverbs is possible. Also see the contrast between Allegedly John will probably resign and *Probably John will allegedly resign. “Evidential” adverbs have to follow “evaluative” adverbs:

(ii) a. Gianni ha purtroppo chiaramente esagerato. ‘G. has unfortunately clearly exaggerated.’ b. *Gianni ha chiaramente purtroppo esagerato.

Siewierska (1992, 418) notes a parallel contrast in English:

(iii) a. Fortunately, he had evidently had his own opinion of the matter. b. *Evidently, he had fortunately had his own opinion of the matter. See also the discussion in chapters 3 and 4 concerning the “evaluative” and “evidential” suffixes present on the verb in a number of languages. 38. Travis (1988, 301) also claims that speaker-oriented adverbs (which, for her, comprise both “pragmatic” adverbs like frankly, and “evaluative” adverbs like unfortunately) have to precede “epistemic,” or S-adverbs, like probably, which in turn have to precede “‘subjectoriented” adverbs like carefully, which have to precede manner adverbs. In observing that

NOTES TO PAGES 12-13 175

“combinations of two speaker-oriented adverbs are often unacceptable,” Jackendoff (1972, 88ff), also notes tangentially that Happily Max has evidently been trying to decide whether to climb the walls contrasts with *Evidently Max has happily been trying to climb the walls. Schlyter (1977, 183) observes the contrast between Malheureusement/évidemment il faut nécessairement y aller ‘Unfortunately/evidently, one has necessarily to go’ and *Nécessairement, il faut malheuresement/évidemment y aller “Necessarily, one has unfortunately/evidently to go.’ 39. By transitivity, “pragmatic” adverbs should likewise precede “modal” adverbs, perhaps, and “subject-oriented” adverbs. And indeed they do. See (ia-b), noted in Schreiber (1972, 330), and (iia—b) (similar facts hold in Italian):

(i) a. Frankly, John obviously/wisely left early. b. *Obviously/wisely, John frankly left early. (1) a. Frankly, John perhaps exaggerated. b. *Perhaps, John frankly exaggerated. 40. This is explicitly suggested in Schlyter (1977, §2.4.2.3). I set aside the question whether speech-time adverbs are moved into, or, more plausibly (see $1.4), “base generated” in, this Topic position. 41. The already cited class of “domain” adverbs (politically, legally, etc.) is plausibly generated in the same Topic position. Like “speech-time” (locative, and other classes of) adverbs, “domain adverbs” also provide the relevant frame within which to evaluate the truth (or appropriateness) of a given speech act. Indeed, it appears that “domain adverbs” have to precede the other sentence adverbs:

(i) a. Politicamente, non c’é francamente da stare allegri. ‘Politically, there are frankly no grounds for being merry.’ b. *Francamente, non c’é politicamente da stare allegri.

(i) a. Politicamente, non c’é purtroppo soluzione. ‘Politically, there is unfortunately no solution.’ b. *Purtroppo, non c’é politicamente soluzione.

(iii) a. Politicamente, non c’é probabilmente nulla da fare. ‘Politically, there is probably nothing to do.’ b. *Probabilmente, non c’é politicamente nulla da fare. (iv) a. Politicamente, non c’é ora nulla da fare. ‘Politically, there is now nothing to do.’ b. *Ora, non c’é politicamente nulla da fare.

(v) a. Politicamente, non c’é forse via d’ uscita. ‘Politically, there is perhaps no way out.’ b. *Forse, non c’é politicamente via d’ uscita. (vi) a. Politicamente, non c’é saggiamente che da attendere. ‘Politically, there isn’t wisely but to wait.’ b. *Saggiamente, non c’é politicamente che da attendere. 42. This recalls Cardinaletti and Starke’s (1994) weak/strong opposition in pronouns, which they in fact explicitly extend to adverbs. I return to the theoretical interpretation of this opposition concerning adverbs in §1.4. Mica is apparently the only one that cannot occur in the postcomplement “space” (see (61a); the same is true of ikke ‘not’ in Norwegian, as @ystein Nilsen points out to me). Apparently, it can in cases such as Gianni, da allora, non ha fumato

176 NOTES TO PAGES 14-18 i suoi sigari mica piu completamente ‘G., since then, has not smoked his cigars not any longer completely’, but the contrast with (61a) suggests that here mica is necessarily in the Spec of piu. 43. De-accented higher (sentence) AdvPs can also be found finally, or interspersed among these de-accented adverbials. 44. Ideally, the theory should leave no choice between these two possibilities. This is in fact what I will argue later. 45. The availability of wh-movement is not simply related to whether or not the adverb allows for degree modification, since probably and (“subject-oriented”) courageously can take degree modifiers (most probably, very courageously), but still resist wh-movement; see (72b and 72c).

46. The same conclusion is suggested by the following contrast in French, noted in Schlyter (1974, 83):

Gj) a. *Energiquement, il a probablement travaillé. ‘Energically, he probably worked’ (cf. Probablement, i] a énergiquement travaillé. ‘Probably, he worked energically.’) b. C’est énergiquement qu’il a probablement travaillé. ‘It’s energically that he probably worked.’

47. No comparable movement of the same AdvPs out of an embedded clause is possible in Italian, which is possibly related to the fact that in Italian, again differently from French (see (i)), these AdvPs cannot be moved to the COMP “space” either:

(i) a. Pour ne pas que tu viennes ... ‘So that you not come...’ b. ??Pour ne rien que tu fasses ... ‘So that you do nothing .. .’ c. Pour ne plus que tu parles ... ‘So that you don’t speak any longer .. .’ 48. When a negative AdvP crosses over a universal quantifier in subject position, however, the scope appears reversed. See // faut que tous ne fument plus ‘It is necessary that all not smoke any longer’ versus II ne faut plus que tous fument ‘It is no longer necessary that all smoke.’ The AdvPs have to respect their “base” relative order in dertved position as well:

(i) a. Il faut toujours tout bien que tu fasses. ‘It is necessary always everything well that you do.’ b. *II faut tout bien toujours que tu fasses. ‘It is necessary everything well always that you do.’

49. The necessity of preventing movement of mal from crossing over toujours has already been recognized by Ruwet (1968, 196). See also Kayne (1975, 27, n. 29). Note that (75) is ungrammatical despite the fact that it can be seen as the combination of the following two well-formed possibilities:

(i) a. Vous avez mal di raccrocher. “You must have hung up badly.’ b. Vous avez di toujours raccrocher. ‘You must have always hung up.’ As long as the order between toujours and mal is respected in either “field” the two can co-occur:

(ii) a. Vous avez di toujours mal raccrocher. “You must have always badly hung up.’ b. Vous avez toujours mal dé raccrocher. “You always badly must have hung up.’ 50. Analogous facts are found in German. Cf (1) and (11), from Bartsch (1976, 229), and Doherty (1985, 115ff), respectively (see also Lang 1979, Pittner 1995, and Pozzobon [998):

NOTES TO PAGES 18-20 177

G) a. Wahrscheinlich kommt Peter oft. ‘Probably comes P. often’ b. *Oft kommt Peter wahrscheinlich. ‘Often comes P. probably.’ (i) a. Wahrscheinlich ist Konrad wirklich verreist. ‘Probably has K. really left.’ b. *Wirklich ist Konrad wahrscheinlich verreist. ‘Really has K. probably left.’ Tarald Taraldsen and @ystein Nilsen (personal communication) inform me that the Scandinavian languages appear to behave similarly. Focalization of the adverb in initial position ameliorates the sentence (as Friederike Moltmann [personal communication] pointed out). This may have the effect of turning the movement of the AdvP into a genuine case of wh-movement (focus movement), making it more similar to the unexceptionable Sehr oft hat Karl wahrscheinlich Marie gesehen ‘Very often has K. probably seen M.’ I take the order reversals via movement of a “lower” adverb to Spec, CP over a “higher” adverb, judged possible in German in Hetland (1992, 162), to be cases of Focus-movement. 51. In a “Split? COMP analysis one might think of the landing site of AdvPs as a Spec position distinct from the one hosting wh-phrases (see Rizzi 1995, Poletto 1997, and references cited there). 52. Koster (1978, 208), in fact, suggests a proto-Relativized Minimality account of these facts in terms of his Locality Condition. 53. As noted in the literature, these adverbs do not enter exactly the same construction, though. Dubuisson and Goldsmith (1976, 109) note that “epistemic modal” adverbs like probablement can be followed by either gue or verb-subject inversion (with no que) (see (i)), while évidemment is compatible only with the former possibility (see (41)):

(i) a. Probablement / Sans doute / Certainement qu’1l viendra. ‘Probably/Undoubtedly/Certainly that he will come.’ b. Probablement / Sans doute / Certainement (*que) viendra-t-il. ‘Probably/Undoubtedly/Certainly will he come.’

Gi) a. Evidemment qu’il viendra. ‘Evidently that he will come.’ b. *Evidemment viendra-t-il. ‘Evidently will he come.’ Peut-étre behaves like the adverbs in (i). Heureusement, like évidemment, can only be followed by gue, while malheureusement is only compatible with subject-verb inversion. It would seem, then, that only negative and non-realis sentence adverbs can (must) trigger subjectverb inversion (J-to-C movement), but further work is needed before the correct generalization can be established. See also Sueur (1978, 237ff), Rochette (1990, n. 15), and Laenzingler (1996b, §2.11.3). In Italian, only forse ‘perhaps’ appears to the left of the complementizer che, in restricted contexts (see Forse che ci aiutera? ‘Is it possible that he will help us?’; Parleremo loro senza forse che si convincano ‘We will speak to them without perhaps that they convince themselves’). 54. See Dik (1975, 104ff). 55. The contrast in (86) is cited in Jackendoff (1972, 64), after an observation of Emonds. The contrast between (87a) and (87b) was pointed out by Tom Roeper (personal communication). But see Keyser and Roeper (1984, n. 7) for some cases where the middle interpretation is apparently licensed even by a preverbal adverb. 56. Evidence for this conclusion comes from a generalization noted in Baker (1989, §11.3.4). Baker observes that the adverbs that can appear immediately after a (stressless) modal or a (stressless) finite form of the auxiliaries have and be “are exactly the same as those that can occur in the position before the finite verb” (p. 291). This suggests that the latter is the

178 NOTES TO PAGES 20-25

basic position available to all verbs, and the former is the same position crossed over only by those few verbs (stressless modals and auxiliaries) that are allowed to raise. See Baker (1971, 1981, 1989), and references cited there, for more careful discussions. 57. An analogous case is provided by the adverb completely in its preverbal and postobject positions: John completely forgot her instructions versus John forgot her instructions completely. The second sentence is ambiguous. It can mean either that John forgot every part of each of her instructions or that they did not occur to him at the appropriate moment (see Moltmann 1995, §6.5). The first has only the latter reading. It is tempting to derive the ambiguity of the second sentence from a structural ambiguity, the reading it shares with John completely forgot her instructions being derived by raising [forgot her instructions] to the left of completely, as proposed here for other cases. 58. If anything, just as with AdvP movement from one “middle field” to the next in French, or with AdvP movement to Spec CP in the Germanic V2 languages, we should expect movement of one AdvP from the lower to the higher “space” to induce a Relativized Minimality violation. But this is not the case.

59. That adverbs may move leftward within larger constituents is also suggested in Sportiche (1993). The derivation in (94), except for the leftward, rather than rightward, movement recalls Schlyter’s (1977, 171) transformation “Rotation,” deriving Pierre a soigneusement chargé la voiture immédiatement from Pierre a immédiatement soigneusement chargé la voiture. 60. In conformity with Kayne (1994), I assume that the movement is to a Spec position (perhaps created by the movement itself, along with a corresponding head). 61. Also note, as observed, that the VP-final “space” has only “lower” AdvPs, and no “higher” (sentence) AdvP. While this cannot easily be seen as a consequence of anything else in the two-“space” analysis, in the one-“space” plus leftward XP-movement alternative, it can plausibly be seen as an indirect effect of the limited possibility of “scrambling” lower portions of the clause up the functional structure of the sentence. If AdvPs in CPs corréspond structurally to A(djective) P(hrase)s in DPs (Cinque 1994), the fact that no attributive AP is found NP-finally (after the N’s complements; see *La ragione del suo rifiuto successiva e principale ‘the reason for his refusal next and main’) could perhaps be attributed to the more limited amount of scrambling of lower portions of the DP up the DP functional structure (in Italian-type languages). 62. If any of these, or any “higher” AdvPs, or extraposed constituents, appear after the focalized mai PIU in (97), they are de-accented (see the following discussion for a speculative analysis).

63. In the case of tutto, it is independently plausible that its position between completamente and bene is a “derived” position, from where tutto, as quantifier, binds a variable in the “base” position of the object. See Kayne (1984, 89ff), Belletti (1990, 77ff), and Cinque (1992a). Its being a bare QP (binding a variable) is plausibly what allows it to cross over AdvPs without inducing a Relativized Minimality violation. 64. This also seems to imply that an approach a la Cardinaletti and Starke (1994) may still be viable for the “derived” and “base” positions of tutto and bene, but should not extend to the case of “lower” AdvPs in pre-VP and VP-final position, which have been argued not to move alone. 65. Note that in each “space” of (99) and (101) the relative ordering of the AdvPs is respected (with sempre > bene; pitt > completamente, pitt > bene). 66. The a priori possible derivation whereby bene is first extracted out of YP, and then YP is moved across mica piu violates Relativized Minimality (bene having crossed over completamente). Also, it seems that scrambling of a constituent XP cannot feed further leftward scrambling of a subconstituent of that XP. For example, if ZP could be extracted from XP in (105f), one would expect *Lui non ha rivisto bene i suoi appunti sempre completamente

NOTES TO PAGES 25-28 179

mica pitt to be possible, contrary to fact. I thank Richard Kayne (personal communication) for raising this question. 67. This is in agreement with Kayne’s (1994) conclusion that elements to the right cannot asymmetrically c-command elements to their left (and take them under their scope). Under that theory, if the scope of a is the domain of those elements (asymmetrically) c-commanded by a, the only way for © to take scope over some element to its left is via “reconstruction” (with that element originating to the right of a, and then moved to its left). 68. He later says that the “difficulty of interpreting [110)b] [106)b] and the questionability of [107)b] seem to point to a difficulty in applying intentionally to a V' with preverbal twice.” Although, as we see directly, the correlation seems correct, why applying intentionally to a V' with preverbal twice” should be difficult remains mysterious in this analysis. 69. For the more precise location of these two positions of twice/often/etc. with respect to other adverb classes, see chapter 4, as well as Obenauer’s (1994, chap. 2, §1) discussion about souvent in French. The two distinct interpretations associated with the two positions are possibly the consequence of applying the same basic meaning to two different portions of the clause. 70. As McCawley (1990, 426) notes, this is particularly clear in such cases as Nineteenth-century composers frequently died young, where the predicate “denotes something an individual can do at most once.” 71. As Luca Dini pointed out (personal communication), the two readings also differ in that the second, but not the first, is compatible with the progressive form in Italian (see (1)). I have no explanation for this contrast:

(i) a. *(Di questi tempi), Gianni sta spesso andando a Roma. “These days, G. is often going to Rome.’ b. (Di questi tempi), Gianni sta andando (spesso) a Roma (spesso). “These days, G. is going to Rome often.’

72. [take the position that the second twice (three times / often, etc.) occupies in (115a) to be the same that it occupies in (115b), the apparent difference being due to the movement of the PP on the door to its left in (115a) (for informational reasons). 73. The appearance of both twice, and twice, disambiguates the sentence, as expected:

(1) a. John twice intentionally knocked twice on the door. b. John twice intentionally knocked on the door twice. Furthermore, the prediction is that a “comma intonation” before the second twice should make the sentence ungrammatical. If “comma intonation” arises as a consequence of XP movement around twice,, we then expect the following to be possible (correctly, it seems):

(ii) a. John intentionally knocked on the door twice, twice. b. John intentionally knocked twice on the door, twice. 74. If twice, were also present, we would then have either (ia) or (ib):

(3) a. John twice knocked on the door twice intentionally. b. John twice knocked twice on the door intentionally. 75. Roberts (1987, 78) also gives the two possibilities in (4), saying that (ia) is ambiguous between (106a) and (106b), while (ib) “shows a preference for the reading with widescope intentionally”:

G) a. John knocked twice on the door intentionally. b. John knocked intentionally twice on the door. For more discussion, see Pesetsky (1989) and Koizuimi (1995, chapter 2).

180 NOTES TO PAGES 28-31

76. See *John will tomorrow/there attend classes; Bill quickly/*with a crash dropped the bananas (Jackendoff 1977, 73), although time adverbs appear possible in journalistic prose (Liliane Haegeman, personal communication). 77. These additional phrases involve, in Chomsky’s examples (131) and (15 1la—b), circumstantial phrases of place, time, manner, as well as extraposed PPs and relative clauses. 78. Though influential, the Davidsonian treatment of adverbial phrases as predicates of

events is not universally accepted. Another influential approach, the “predicate-modifier’ approach, stemming from Montague (1974), considers them “as making a more complex predicate out of a simpler one” (Cresswell 1985, 4), essentially treating all adverbial elements as operators. See Thomason and Stalnaker (1973), Cresswell (1985), and Parsons (1990), for discussion and references. 79, I take Jackendoff’s (1972, 59, 82) class of “auxiliary position” adverbs (merely, simply, etc.) to be “focusing” adverbs. This may allow an account of certain puzzling facts noted in Lonzi (1990) for Italian. While Ha meramente annuito ‘He has merely nodded’ is grammatical, and Meramente annuire (é inutile) ‘Merely to nod (is useless)’ is also marginally possible, *Annui meramente ‘He nodded merely’ is not. Meramente patterns with perfino ‘even’, rather than with solo/solamente/soltanto, in not allowing its complement to raise past

it. See Ha perfino/solo/ ... annuito; Perfino/solo/ ... annuire (é inutile); Annui solo/ solamente/soltanto/*perfino. 80. This may suggest that finite verbs, as opposed to finite auxiliaries, raise only to a head lower than the “higher” adverbs (in fact, lower than habitual adverbs). See Belletti (1990) for remarks in this direction. An exception to this generalization is provided by the subclass of (realis) mood adverbs comprising sicuramente / di sicuro ‘surely’, certamente / di certo ‘certainly’, senz’altro ‘surely’, and so on (but not senza dubbio ‘undoubtedly’). See (1):

(1) Gtanni lo merita sicuramente/ di sicuro / etc. / ??senza dubbio. ‘G, deserves it surely/undoubtedly.’ Perhaps (i) involves a “focusing” use of sicuramente, with sicuramente patterning like solo, rather than perfino (..e., allowing its complement to raise past it). Some evidence in this direction 1s provided by the acceptability of A NOI SICURAMENTE, lo dara ‘To us surely (focus), he will give it’ (alongside SICURAMENTE A NOL lo dara) versus ??A NOI SENZA DUBBIO, lo dara ‘To us undoubtedly (focus), he will give it’ (see SENZA DUBBIO A NOL, lo dara). Also see *A NOI PROBABILMENTE/FORTUNATAMENTE, lo dara ‘To us probably/luckily (focus), he will give it’ versus PROBABILMENTE/FORTUNATAMENTE A NOL lo dara. 81. Belletti proposes that in such cases these adverbs (probabilmente in her examples) are directly adjoined to the following constituent. A similar suggestion is made in McCloskey (1992, n. 2), where sentence adverbs in this usage are explicitly equated to only and even. See also Laenzingler (1996a, n. 1). 82. Even clearer evidence is provided by their fronting together with another constituent in the first position of root clauses in V/2 languages. See the German example in (1) (from Bartsch 1976):

G) Vermutlich das Buch hat Peter gestohlen. ‘Presumably the book has P. stolen.’ 83. Versus *C’est heureusement que Paul a vendu sa voiture ‘It’s luckily that P. sold his car’. Contrasts such as these were noted by Robert Martin (see Nglke 1990, 119). 84. That spesso/often/etc. do not admit of focusing usages is shown by the contrast between (128b) and (1) (see Schlyter 1977, 179ff): (1) *E’ spesso per questa ragione che s1 arrabbia. ‘It is often for this reason that he gets angry.’

NOTES TO PAGES 31-35 181

PPs, but no extended projection of VP (except CP), can occupy the focus position of a cleft. 85. This is not possible with perfino ‘even’. See (1):

G) a. Perfino a suo FIGLIO lo dira. ‘Even to his son he will say that.’ b. *A suo figlio PERFINO lo dira. ‘To his son even he will say that.’ 86. Also not investigated here are the restrictions on what constituents each adverb can attach to, when used as a “focusing” adverb. 87. As observed in Browning (1996, 238, n. 2), a parenthetical (or “interjection” intonation of the adverb (“which signals that it is not actually part of the syntactic representation”) even allows apparent violations of the adjacency requirement on verbs and objects in English. See (i):

(i) a. *He bought unfortunately a Yugo. b. He bought, unfortunately!, a Yugo. 88. By transitivity, an adverb like probably should precede an adverb like perhaps or (almost) certainly. While this appears to be true for probably and perhaps (which do show this order, if co-occurring at all: Probably he will perhaps try again versus *? Perhaps he will probably try again), probably and (almost) certainly, according to Richard Kayne (personal communication), seem to admit either order, which, in the present context, may be problematic. 89. The sentences are preferable with no longer:

(i) a. John no longer always wins his games. b. *John always no longer wins his games. Richard Kayne tells me that for him the order always no longer is not unacceptable (e.g., He always no longer wants to go when you ask him). It is possible that always, like other quantificational adverbs, is generable in two distinct quantificational spaces: one to the left, and one to the right of no longer. See the last paragraph of n. 12, and chapter 4. The reason this is not visible in the Italian (see (141)) may be related to preverbai negation. No more/(not ...) any more minimally differs from no longerXnot ...) any longer in being compatible with only one of the two possibilities. See (41)—(i11):

(ii) a. *John no more always wins his games. b. John always wins his games no more. (possible in British but not in American English)

(iii) a. *John doesn’t any more always win his games. b. John doesn’t always win his games any more. In (my interpretation of) Cardinaletti and Starke’s terms, no more / (not) any more are only strong adverbs, requiring leftward movement of the constituent following them. 90. For the sentences and judgments reported here, I am indebted to @ystein Nilsen. For more detailed discussion, see Nilsen (1997, 1998). 91. As Mystein Nilsen points out (personal communication), speech—time adverbs like nd ‘now’ can also precede speech-act adverbs, as was noted for Italian (Per forlater nd erlig talt heldigvis sannsynligvis selskapet ‘Peter leaves now honestly spoken fortunately probably the party’). 92. “Not yet’ is realized either as ikke enda (‘not still’) or enda ikke. There is some indication that the latter, as opposed to the former, is not a single constituent. As @ystein Nilsen pointed out to me, enda and ikke can be separated by other elements. See (i), possible with stress on tkke, and igadr:

182 NOTES TO PAGES 35-40

(i) Igar forsto enda Per det ikke helt godt. yesterday understood still P. that not completely well. “Yesterday P. didn’t understand that completely well yet.’ The fact that nothing can interrupt the sequence ikke enda, and that it, but not generally enda ikke, can constitute an isolated answer, suggests that ikke enda is a constituent. 93. As Mystein Nilsen notes, the sentential negation ikke can (and must) precede allerede (See Per har vel ikke allerede/*allerede ikke vunnet! ‘P. has really not already/already not won’; and the order mica gid versus * gid mica tn Italian, discussed in §1.1); however, it cannot if lenger follows allerede (*Per elsker ikke allerede lenger sin kone ‘P. loves not already any longer his wife’), which might suggest that the ikke of ikke lenger is not the sentential ikke, but a ikke directly modifying lenger (see Nilsen 1998). 94. The relevant sentences and judgments were kindly provided by Nedzad Leko and Ljiljana Progovac. 95. MoZda ‘perhaps’ is not felt possible with vjerovatno ‘probably’. 96. Nedzad Leko informs me that this sentence, if it receives any interpretation at all, forces a manner (not a subject-oriented) reading of inteligentno. 97. For certain speakers these two adverbs cannot occur together (Ljiljana Progovac, personal communication).

98. For Ljiljana Progovac, (168a) is only possible in the reading in which potpuno directly modifies dobro. 99, Tam grateful to Ur Shlonsky for providing the relevant sentences and judgments on Hebrew. For an overview of a comparable sequencing of adverbs in another Semitic language (Arabic), see Fassi Fehri (1997).

100. Ur Shlonsky points out that (171b) is acceptable if be’emet is interpreted as ‘really’ rather than as a speech-act adverb. 101. Kanir’e ‘probably’ can apparently precede le mazal-a ‘luckily’ if it is in the COMP space (Kanir’e Se le mazal-a Rina tenaceax ‘Apparently that luckily R. will win’). See n. 50, and relative text. 102. The sequence ‘ulay kanir’e is grammatical with ‘ulay interpreted as an optative particle (Ur Shlonsky, personal communication), which I take to involve a shift to the speech act adverb class. 103. “No longer” is expressed by the combination “already not” (as in other languages; see van der Auwera 1991):

Gj) a. Hu kvar me a§Sen. ‘He already smokes.’ (i) Hu kvar (bixlal) lo ohev ota. he already (at all) not loves her. ‘He doesn’t any longer (at all) love her.’ 104, The (finite) verb has to precede the adverb ‘well’ (see *Hu heitev hevin ‘et dvarexa ‘He well understands your words’), and in general also precedes the “completely” adverb (though it may marginally follow it in particular contexts; Ur Shlonsky, personal com—

munication). For the similar situation of Italian, Norwegian, and related languages, see chapter 5. 105. The examples come from Camporese (1996) and from personal communications from James Huang and Liejiong Xu. The source is indicated in parentheses as (C), (H), and (X), respectively. For the order of other “lower” adverbs and their position with respect to the verb, see Tsai (1995). 106. The same contrast is noted in Camporese (1996, §2.3), who reports the following examples:

NOTES TO PAGES 40-45 183

(i) a. Buxing, xianran wo dui nimen sangshi xinxin. unfortunately evidently I in you lost faith. ‘Unfortunately I have evidently lost my faith in you.’ b. *Xianran, buxing wo dui nimen sangshi xinxin. 107. Xianzai ‘now’ too precedes mingzhide ‘wisely’. See G), from Camporese (1996, $2.3):

Gi) a. Taxianzai mingzhide rangbu le. he now wisely withdraw PERF ‘He has now wisely withdrawn.’ b. *Ta mingzhide xianzai rangbu le. 108. For Liejiong Xu, both orders of mingzhide ‘wisely’ and yiban ‘usually’ appear possible. 109. Given (189) and (190), by transitivity, we expect yiban ‘usually’ to necessarily precede yijing ‘already’, which is the case, as noted in Camporese (1996, §2.3):

(i) a. Ba dian, ta yiban yijing kaishi gongzuo le. at 3, he usually already has begun work PERF b. *Ba dian, ta yijing yiban kaishi gongzuo le. 110. Note that given the order yijing ‘already’ > bu-zai ‘no longer’ > zongshi ‘always’ > yizhi ‘continuously’, we expect yijing ‘already’ to precede yizhi ‘continuously’. This is what we find. See (1), from Camporese (1996, §2.3):

(i) a. Wo yijing yizhi pao le wushi gongli, keshi hai bu lei. I already continuously run Asp fifty km, but still not be tired. ‘T have already run for fifty km, but I’m not yet tired.’ b. *Wo yizhi yijing pao le wushi gongli, keshi hai bu lei.

111, The Albanian data and judgments were kindly provided by Dalina Kallulli. 112. Rackowski considers the adverbs which are crossed over to be heads, while the ones preceding the V to be XPs, but her account carries over, with minor adjustments, to an analysis in which all adverbs are XPs (in Spec), as assumed here. She also considers, correctly, I think, the absolute final position of speech act adverbs to derive from a lefthand

movement of the VO...S sequence across it.

Chapter 2 For comments on this chapter, I thank Adriana Belletti, Paola Beninca’, Richard Kayne, and Luigi Rizzi. 1. That AdvPs should be specifiers is not only suggested in Kayne (1993a, n. 42; 1994, 137, n. 30), but also in Laenzingler (1993, 1996a,b) and Alexiadou (1994a,b; 1997), among others, as well as in earlier unpublished work of mine (Cinque 1992b, discussed in Belletti 1994a and Alexiadou 1994a,b). Laenzingler’s idea that AdvPs are licensed via a Spec/head relation with a head (his Adverb Criterion) bears some similarity to the idea presented here, though the specifics of his analysis are rather different, and, in fact, incompatible with the conclusion defended here of a unique (left-hand) specifier per head. Also see Rivero (1992) and Ojea Lépez (1994) for the suggestion that aspectual and temporal adverbs are adjuncts of Aspect and Tense projections. 2. I will return to the relevance of these facts for other versions of X-bar theory (such as that in Chomsky 1995), which allow for multiple specifiers in between two heads.

184 NOTES TO PAGES 46-51

3, This raises the interesting question of what it is that triggers the obligatory and the optional instances of (active) past participle movement. I will come back to (some speculations about) such questions. 4. The possibility for tottu to precede an (active) past participle is attributed more generally to Sardinian in Jones (1993, 209). (Logudorese) Sardinian appears to be just like Italian for all higher (apparently optional) movements of the past participle (Patrizia Ruggiu, personal communication). 5. Note that this analysis is compatible with taking (certain) AdvPs to move, as long as they end up forming a fixed sequence within a particular portion, or “space,” of the clause, as discussed in the previous chapter (§1.4). 6. It could be claimed that the same ordering principle needed to determine the correct order of adjunction within position A, and position B, with mica > pii, is sufficient to account for the ungrammaticality of (9b); at least if the principle is abstract enough (possibly semantic) to hold across positions A and B as well. But even so, Alternative If compares unfavorably with Alternative I, which needs no separate ordering principle for AdvPs as specifiers, as their ordering can plausibly be identified with the independent ordering of the heads (via Spec/head agreement), as I will claim later. 7. Neither can (9b) be derived from (10a) by moving [piu mangiato] across mica. As noted in chapter 1, n. 42, mica is the only one of the “lower” adverbs that cannot occur focalized in VP-final position. I thank Rita Manzini for bringing this to my attention. 8. If AdvPs have to be in the “checking domain” of a head, because of the special relation they have with it, then they must be in Spec positions also in a system like that of Chomsky (1995) (which also allows for adjunctions). The reason is that only specifiers, and no adjuncts, fall inside the “checking domain” of a head (see Chomky 1995, 319 and his n. 47). As noted, the evidence discussed here also favors single-specifier, over multiple-specifier, phrases. If the latter were possible hosts of AdvPs, one would then expect heads to be barred in between AdvPs (and argument DPs, as well). Taking the V to raise in between the AdvPs found in multiple Spec’s only in the phonological component (in analogy with what is proposed in Chomsky 1995, 368) may not be appropriate here, as the different location of the past participle among the AdvPs has often semantic consequences (see chapter 4, §4.24), Also see the contrast between (ia) and (ib):

G) a. Gianni ha sempre avuto 1 capelli lunghi. ‘G. has always had long hair.’ b. Gianni ha avuto sempre 1 capelli lunghi. Example (ia) with the past participle to the right of sempre implies that Gianni still has long hair (a non-“perfect” interpretation), whereas (ib) is compatible with both a “perfect” and a non-“perfect” interpretation. 9. In (15)-(17), I have abstracted from the position of the subject to avoid introducing too many variables at a time. I will come back to this question in chapter 5, after discussing the position of AgrPs in the clause.

10. The pragmatics of this use are slightly different from those of postverbal mica. Preverbal mica can be used only as the denial of someone’s positive assertion (not just as the denial of a generic expectation). 11. Note that mica can precede a subject, but in this case it contrastively negates just the subject: Mica Gianni ci telefonerda; ci telefonerd Mario ‘Not G. will call us; Mario will.’ 12. Example (22b) is grammatical in the irrelevant reading in which mica modifies the matrix clause. 13. The movement analysis proves, in fact, untenable, as it leads to a paradox. Consider (i) and (il):

NOTES TO PAGES 51-52 185

(i) a. Gianni non é mica ripartito. ‘G. hasn’t not left.’ b. Gianni mica é ripartito. ‘G. not has left.’ (ii) a. Probabilmente Gianni é ripartito. ‘Probably G. has left.’ b. Gianni probabilmente é ripartito. ‘G, probably has left.’ c. Gianni é probabilmente ripartito. ‘G. has probably left.’ Suppose it was adverbs that moved. If raising, but no lowering, is admitted, examples (ia) and (ic) should represent the “base generated” position of mica and probabilmente, with (ib) derived by fronting mica, and (iia—b) derived by fronting probabilmente (example (1b), perhaps further involving topicalization of the subject; Belletti 1994b). But, then, a perfectly acceptable sentence such as (iii) should be ungrammatical as it would involve a violation of Relativized Minimality (due to probabilmente having eventually crossed over mica):

(i1i)Probabilmente Gianni mica é ripartito. ‘Probably G. not has left.’ If, on the other hand, an AdvP were allowed to cross over another, in disdain of Relativized

Minimality, we should expect (ivb) to be possible, along (iva), contrary to fact:

(iv) a. Tu francamente avresti probabilmente dovuto fermarti. “You frankly would probably have had to stop.’ b. *Probabilmente tu francamente avresti dovuto fermarti. ‘Probably you frankly would have had to stop.’

A movement analysis thus leaves no way out. For further arguments against a movement analysis of adverbs, see T. B. Ernst (1991).

14. The ancora of non... ancora ‘not yet’ can also, apparently, occupy a distinct preverbal position preceding non (see Belletti 1990, 127, n. 7): Ancora non V’ha visto ‘Yet not he has seen it’. This position, however, can once again be shown not to be a function of ancora moving to a higher position. Rather, it seems to be due to the finite verb stopping in a lower head position as its relative order with respect to the other classes of AdvPs remains the same, apparently.

GQ) a. *Ancora non l’ha mica visto. “Yet not he has not seen it.’ b. Mica ancora Vha visto. ‘Not yet he has seen it.’

(ii) a. *Ancora non l’ha probabilmente visto. “Yet not he has probably seen it.’ b. Probabilmente ancora non 1’ha visto. ‘Probably yet not he has seen it’ For further evidence that finite verbs in Italian can stop in various lower heads to the right of (most) “lower” AdvPs, see chapter 5.

Chapter 3 The material discussed in this chapter was presented in lectures at the Universities of McGill, Lund, Groningen, Amsterdam, and Rome. I am indebted to those audiences for comments and criticism, and in particular to Mark Baker, Christer Platzack, Lisa Travis, Sten Vikner, and Jan Wouter Zwart. I also thank Enoch Aboh, Adriana Belletti, Paola Beninca’, RoseMarie Déchaine, Michel DeGraff, Richard Kayne, Jaklin Kornfilt, Victor Manfredi, Cecilia Poletto, and Luigi Rizzi for comments to an earlier version of the chapter.

1. With “closing’/“nonclosing” suffixes I refer, after Nida (1974, 85), to those suffixes that close off/do not close off the word, thus disallowing/allowing further suffixation (see also n. 43).

186 NOTES TO PAGES 52-54

2. In the early work, the inflected word was taken to be literally built up in the syntax via successive adjunctions. If words with inflectional suffixes are more plausibly taken to come already formed from the lexicon, with inflectional features to be “checked” during the syntactic derivation (Chomsky 1995, chap. 3, 195ff), Baker’s Mirror Principle retains its validity if the order in which the features are checked reflects the order in which they were assigned to the word during the morphological derivation (which remains to be determined precisely. See Speas 1991a, 409ff., and Chomsky 1995, 195ff), The fact that inner suffixes are checked before outer suffixes possibly follows from the principle of the cycle (applied to morphological constituents). An advantage of the “checking” approach is that it makes it more natural that derivational suffixes (which even more plausibly originate in the lexicon) should fall under the Mirror Principle. In fact, in many languages “lower” functional heads are often realized with derivational rather than inflectional (or agglutinating) suffixes. As derivational suffixes are closer to the stem than inflectional (or agglutinating) ones, certain predictions follow for the order of the functional heads to which they correspond, an issue to which I return in §3.5. 3. Yoon (1990) treats declarative -fa as a kind of complementizer, while Ahn and Yoon (1989), Bhatt and Yoon (1991), and Cho (1994) treat it as a sentence mood verbal suffix, followed, when embedded, by the general subordinator -ko. I leave open here the question whether these speech act moods should be identified with the head of Rizzi’s (1997) ForceP, within the complementizer “space,” returning to the question in chapter 4 (§4.4). 4. See also Aksu-Kog and Slobin (1985, 84), who claim that the sequence of suffixes in Turkish conforms to the order (stem-)reflexive-reciprocal-causative-passive-potentialnegation-necessitative-tense-conditional-person-number. 5. Although the modal suffix cannot be directly affixed on to the progressive aspect suffix (requiring instead the insertion of an auxiliary), the relative order of the corresponding heads is arguably Mod > Aspprogressive: S€€ (66C) in the text.

6. For an analogous conclusion based on Romance, see Picallo (1985, 232ff.; 1990, 287ff.). In the American varieties discussed by Boertien and Battistella, the first modal (might) shows an exceptional pattern with respect to negation, inversion, and tag-formation, which suggests that it may have been reanalyzed as an adverb (= maybe), though problems remain (for discussion, see Roberts 1993, 333, n. 5). 7. Another (Papuan) language apparently showing epistemic modal suffixes outside the (future) tense suffix (and inside speech act suffixes) is Tauya:

(i) Ni-?e-rafo-?a (MacDonald 1990, 207) eat-3sgFUT-DUB-DECL ‘Maybe he’ II eat.’ In contrast to MacDonald (1990), I have rendered the suffix - 7a as DECL(arative) rather than

IND(icative), as it alternates with such other speech act suffixes as “interrogative,” “exclamative,” “imperative,” and so on. A suffix outside of the future tense suffix, rendered as “probably”, is also found in Evenki (Altaic):

Gi) Asatkan duku-d’anga bi-ne-n. (Nedyalkov 1997, 266) girl write-FUT be-PROBAB-3sg ‘Probably this girl will be able to write (it).’ A “probable mood” suffix also occurs to the right of present and past tense suffixes in Oksapmin, of New Guinea (iii), and in the Sino-Tibetan language Garo (see (iv)):

(iii) Go haanip di-kin-o poripti. (Lawrence 1972, 33) you person eat.PAST-PROBABLE-quotation marker say then ‘You probably attacked the person” we say, then...’

NOTES TO PAGES 54-55 187

(iv) Anti-ci re’an-aha-kon. (Bybee 1985, 180) market-to go-PAST-PROBABLE ‘| think he went to the market.’ 8. Asa matter of fact, there is evidence even internal to Korean for the order Mod, stemsc > T > Mod,,o. Alongside (3) (showing the order Mod, pistemc > T), one also finds cases such as (i) (from Yoon 1990, 358), which arguably suggest the order T > Modo:

G) Chelswu-nun keki ka-ci mos ha-ess-ta. C. TOP there go-ci cannot do-PAST-DECL ‘C. could not go there.’ Tn G), the verb (with the nominalizer -c7) is followed by a (negative) root modal mos, in turn followed by the Tense and “speech-act” Mood suffixes attached to the dummy verb ha ‘do’ (the Korean equivalent of do-support in English). This “analytic” verbal sequence can be accounted for if mos is taken to be a free head morpheme blocking adjunction of the lexical verb ka ‘go’, and its further movement to Tense and Mood (thus requiring insertion of ha), and if the VP is raised past mos and the resulting constituent is raised past ha-ess-ta, in the way characteristic of OV languages as analyzed in Kayne (1994, §5.5). If this interpretation of the facts 1s correct, root Modality, then, is lower than (Tense and) epistemic Modality even in Korean. 9, “Le indicates termination rather than completion” (Smith 1987, 96)—that is, stopping an activity with a natural end point at an arbitrary point. Instead, completion, marked by -wan-, means stopping at the natural end point of that activity. As Smith notes, (11) without -wan- (but not (11)) can be followed by a sentence like jintian hai zai xie ‘today still PROG write (and today he is still writing it)’. For further discussion, see chapter 4. 10. For the order perfect aspect > progressive aspect see (1), from (Huanuco) Quechua (Landerman and Frantz 1972, 147) and the converging evidence coming from the order of the corresponding free morphemes in “head-initial” languages discussed in the next section (also see the case of Imbabura Quechua in Appendix 2):

(i) Juan taki-yka-sqa. J. sing-PROG-PERF ‘Juan has been singing.’ 11. Similarly, Foley and Van Valin (1984) claim that in languages with aspect and tense as separate inflectional categories, “aspect occurs closer to the verb stem, the nucleus, than does tense... . We know of no cases of the inverse order of the two inflectional categories, in which tense is closer to the stem than aspect” (p. 210). The same conclusion (limited to polysynthetic languages) is found in Baker (1995): “Aspect morphemes in the polysynthetic languages consistently appear as suffixes placed between the verb root and the tense/mood inflection, a position that 1s explained if they head a phrase that is lower than IP but higher than the lowest VP” (p. 231). Blansitt (1975) observes that “when the progressive marker is a suffix and any tense, mode, or person markers are also suffixes, all the latter follow the progressive marker” (p. 10). In Payne (1985) it is noted that in the Tungus languages “the negative auxiliary carries tense, mood, person and number markings, whereas other verbal categories such as aspect and voice are indicated by the lexical verb” (p. 215), which consistently points to the latter being lower than the former. Also see Ouhalla (1991, 76) and Speas (1991a, 401 ff). Aspects (frequentative, perfect, durative, predispositional, inchoative, and so on) are also lower than tense in American and Swedish Sign Languages (see Klima and Bellugi 1979, 243ff; Aarons et al. 1992, 1995; Bergman and Dahl 1994). 12. Under status they also collapse a set of notions (“obligatory,” “acceptable,” etc.), which should rather be assigned to Modality, it seems (with no ensuing inconsistency, as far as I can see).

188 NOTES TO PAGES 56-57

13. These suffixes qualify the speaker’s commitment about the truth of the proposition on the basis of the evidence he/she has: direct (visual, auditory, etc.), or indirect (hearsay, etc.). They are a common feature of verbal inflection in American Indian languages, but they are also attested in other language families: Baltic, and South Slavic languages, Albanian, Altaic (Turkish, Japanese, Korean) and other South Asian languages; and Australian and Bantu languages, for example. See Givén (1982a), Chafe and Nichols (1986), Hoff (1986), Palmer (1986), Willet (1988), and Dendale and Tasmowsky (1994). Other languages (such as Basque; see Ortiz de Urbina 1989, 129) use instead evidential particles. In languages lacking evidential suffixes or particles, other ways exist to express the same notions. [ return to these questions in chapter 4. 14. Foley and Van Valin (1984, 223ff) try to derive this order from their three-layered analysis of the clause into a nucleus (the predication), a core (embedding the nucleus and its arguments), and a periphery (embedding both), by viewing Modality as a core operator, and Status, Tense, Evidentiality, and [llocutionary force as periphery operators (Aspect being a nucleus operator). They represent the layered structure of the clause with operators as in (i) (their Fig. 9, p. 224):

G) ULLOC FORCE (EVID (TENSE (STATUS (L..In (MOD [NP (NP) (ASPECT [Predicate]))])))))) A similar model has been developed by Simon Dik and others, also working within a functionalist theory of the sentence. See Dik (1989), Hengeveld (1989, 1990), Cuvalay (1995), and, for a comparison between the two approaches, Van Valin (1990). An insight of Dik’s functionalist model, which is particularly relevant here, is the assignment of adverbial classes to different layers, in association with different functional operators. See in particular Dik et al. (1990), Nuyts (1993), and Hengeveld (forthcoming a). Foley and Van Valin (1984, 223), citing Johanna Nichols, note that a comparable universal order of ver-

bal affixes (voice, aspect, tense (of aspect), mode (tense of mode)) is also proposed in Tesniére (1939). 15. From the present perspective it is of no significance that possibly no language displays the whole array of such suffixes, nor that languages which have specific subsequences of such suffixes do not allow them to occur all together, although the reasons for that merit to be understood. For further discussion, see the next chapters. Extreme polysynthetic languages such as West Greenlandic Eskimo appear essentially to conform to the relative ordering in (18). Fortescue (1980, 261ff) indicates the following as the gross order of suffix classes, with those to the left being in the semantic scope of those to

the right: V ... manner-degree-aspect-tense-epistemic modality-affixes of subjective coloration-inflection (marking speech act mood, person and number). See also Fortescue (1984, §2.1.3). On the case of Navajo, see §3.5 here. 16. Ouhalla (1991, 77ff) and Poletto (1993) take these additional verbs (“auxiliaries”’) to be directly inserted under distinct functional heads rather than heading a VP of their own. See also Drijkoningen (1986). An advantage of directly generating auxiliaries under functional heads would seem to be the possibility of assigning one and the same structure to sentences with compound and simple verb forms. See chapter 6 here (§6.1) for further discussion. Auxiliary insertion may also occur in agglutinative languages when a morpheme does not admit (further) affixation. An example of this (the negative root modal mos of Korean) was discussed in n. 8. Additional cases are discussed in §3.4. 17. In other words, “the order of auxiliaries is a direct reflex of the order of the affixes” (Drijkoningen 1986, 58). This is true even if (some of) the V+ suffix combinations undergo head-to-head movement. As Relativized Minimality bars raising a head over another head (or its trace), the relative positioning of two heads will still reflect the relative order of the

NOTES TO PAGES 57-59 189

positions in which they were inserted. In chapter 1, we have in fact noted the possible leftward scrambling of a single lower portion of the clause even in Italian (more generally, in “head-initial” languages). What is apparently missing in these languages, as noted, is the possibility of successive leftward movements. 18. Examples (20a—b) are drawn from Zagona (1988, 5). The order in (21) is already recognized in Chomsky (1957). Foley and Van Valin (1984, 225) (see also Bybee 1985, 197) notice the “perfect match-up,” in terms of closeness to the verb, of the English auxiliary order and the universal template of the operators arrived at on the basis of the relative order of suffixes. The question of auxiliary choice (have, be, etc.) for each suffix, within and across languages, is orthogonal to present concerns and is not addressed here. For discussion, I refer to Kayne (1993b). 19. This order is compatible with the order of functional heads arrived at on the basis of the order of suffixes. 20. Basque is another case, although the presence of aspect, mood, and modal particles makes it a mixed system. See (58) in the text and also Laka (1990, 18ff). Apparently, negation (in root clauses) prevents the complement of the auxiliary from raising leftward. See (ia—b),

from Laka (1990, 25ff):

(4) a. *Etxea erori ez da. house-the fallen not has ‘The house has not fallen.’ b. Ez da etxea erori. not has house-the fallen ‘The house has not fallen.’ 21. The complements of the lexical verb instead possibly raise separately to different AgrPs, which leaves the V free to move to a functional head. 22. Although categorially both are heads, I take particles to differ from auxiliaries in being invariant (inflectionless) free morphemes. Particles may also differ from auxiliaries in being less prone to movement (perhaps as a consequence of their being poorer in features). But see the evidence for the movement of to in English, discussed in Bernstein (1994). Baltin (1995) discusses similar facts, without reaching that conclusion. 23. This is at least true of the so-called basilect, the variety least influenced by the superstrate language (English, French, etc.), the acrolect being the one most influenced. The mesolect is an intermediate variety, which can show certain features of the superstrate language. So, for example, Rottet (1992) shows that in the Louisiana Creole basilect verbs do not move (see DeGraff 1994 for a similar claim about Haitian Creole), while they move at least partially in the corresponding mesolect. Analogously, in the Guyanese Creole basilect verbs are completely uninflected while some inflection is found in the corresponding mesolect (see Gibson 1986 and the discussion in this chapter which is based on it). Only basilects will be considered here. 24. Gibson translates it as already with stative predicates (Meeri don gat di buk ‘Mary already has the book’ (1986, 584); Shi don noo wu fu duu ‘She already knows what to do’ (1992, 60), and as present perfect with nonstative predicates (Wen you get de, shi gu/gon don sing “When you get there, she will have sung’ 1992, 60). On T(Anterior) (T3), see the discussion in chapter 4. 25. The durative aspect particle de expresses the fact that “a given situation is conceived of as lasting for a certain period of time” (Gibson 1992, 57) (whence, presumably, its incompatibility with stative verbs). In the mesolectal variety, habitual aspect is expressed with the particle (or auxiliary) doz, and progressive aspect with the suffix -ing:

Gi) Shi doz aalweez de singing. (see Gibson 1986, 572ff) she HAB always DUR sing-PROG ‘She is usually always singing.’

190 NOTES TO PAGES 59-61

26. Don is apparently the only particle that can follow the verb (and its complements):

i) Shi get tuu pikni don. (= (b) of Gibson 1986, n. 7) ‘She already has two children.’ In a way reminiscent of the observed reversals of adverb orders in chapter 1, [ take (i) to involve the (optional) raising of [get tuu piknii| to a Spec preceding don. For the analogous behavior of Sea Island Creole done and of the Sranan particle kba, see Cunningham (1992, 52) and Seuren (1983, n. 5), Winford (1998a,b), respectively. Leblanc (1989, 60ff), following a suggestion of Lumsden’s, reaches the same conclusion for sentence final deja in Haitian Creole.

27. See the similar case of the Melanesian language Anejom, (25b) in Appendix 2 and the mirror-image order of Future and Past tense suffixes in Turkish (see (5b), and Lezgian (Caucasian) and Mongolian (see (6b) and (3b) in Appendix 2), which provide evidence for the same order: T(Past) > T(Future). See also the discussion in chapter 4 (§4.2). In this connection, it is tempting to interpret “conditional” woudd in English as the result of will (future) raising to T(Past). The complex (“conditional”) preverbal particle yi-a (lit. FUT-PAST) of Koyo (a Kru language of West Africa) is not an exception to the order T(Past) > T(Future) if taken to derive by adjunction of the lower T(Future) particle yi to the higher T(Past) particle ya. See (ia—c), from Kokora (1976, 65):

(i) a. Abi yi (ta) sakaa-a lu. A. FUT (NEG) rice-the eat “A. will (not) eat the rice.’ b. Dofi ya duN mo. D. PAST town-in go. ‘D. went to town.’ c. N ka-a leyere lu O yi-a saka lu. I COND-PAST wealth eat he FUT-PAST rice eat. ‘If I were rich he would eat rice.’ Another case plausibly derived through the incorporation of a lower to a higher particle is what Mosel and Hovdhaugen (1992, 354) refer to as the “Progressive-Perfect” compound particle ‘olo’ua, in (head-initial) Samoan (Austronesian), which combines the Perfect particle ua and the Progressive particle ‘olo’o (or 0), reversing the expected order. See (26c) in Appendix 2. In th Niger-Congo language Kana, the progressive particle appears prefixed to the past tense particle (cf. Ikoro 1996, 173f), also suggesting incorporation. 28. Example (29) also shows two root modals (of ability, and permission, respectively) following the past tense particle, similarly to what was found in Korean and Turkish, as already discussed. 29. Also, it may be possible to account for the multiple ambiguity of (1) (see Seuren 1981, 1051; 1983, 228) and for the “future in the past” use of o in (ii), mentioned in Winford (1998a):

(i) Mi ben sabi. I PAST know. (a) ‘I knew’, (b) ‘I have known’, (c) ‘I used to know’, (d) ‘I had known.’ Gi) A ben taigi mi a o kon na fesisey baka. he PAST tell me he FUT come LOC frontside back ‘He told me he would come to the front again.’

Statement (ia) would represent mi ben, sabi; (1b) mi ben, sabi; (1d) mi ben; ben, > 0 sabi; Gi) ben, > 00. 30. Retrospective “aspect” is sometimes called “immediate past” (see Corne 1977,111; Dahl 1985, 127; and the discussion in chapter 4, §4.19). Example (ia), from Sylvain (1979, 97), and (ib), from L. H. Desouvrey (personal communication), suggest a certain relative order of the aspectual particles (see (i1)):

NOTES TO PAGES 61-62 191

(i) a. L’té kon fék lévé... 3sg PAST HAB RETRO get up... ‘Il venait d’ordinaire de se lever .. .” (‘He used to have just got up...) b. Jan fék ap kuri. J. RETRO PROG run ‘J. has just been running.’

(ii) a. konn > fék (sét) > ap b. ASPhabitual > ASP retrospective > ASP progressive

Other aspectual particles appear to exist: fin, possibly marking “inceptive” aspect (see example (iiia—b)), apparently intervening between fék (sdf) and ap; and the “prospective” (see n. 39) particle (a)pral, which in Haitian, though not in Fongbe (and Gungbe), is in complementary distribution with the progressive particle (Lefebvre 1995, 165ff; 1996, §§1.1, 1.2). This might suggest that (a)pral in fact incorporates the progressive particle ap(r), which 1s consistent with the fact that in both Fongbe and Gungbe the prospective particle cannot be used without the progressive one.

(iii) a. Vé3 6, Mari konn fék fin ap leve. (L. H. Desouvrey, personal communication) around 3 o’clock, M. HAB RETRO INCEPT PROG get up ‘Around 3 o’clock, M. usually just began to be getting up.’ b. Li t’a-mét-fek-sot-fin-pra-kuri, ... (Sylvain 1979, 101) 3sg PAST FUT MOD RETRO INCEPT get running, . . . ‘Even if he had just begun to run, ... Most of these aspectual particles are related to full verbs (see Magloire-Holly 1982, 94), but they can be distinguished from the latter in terms of meaning and position. So, for example, fin as a verb meaning ‘finish’ follows the progressive particle ap (Jan ap fin kuri °J.1is finishing his running’), while it precedes ap, as noted, in its use as an (inceptive) aspect particle (Jan fin ap kuri, with the meaning ‘Finally, J. started running’). I thank L. H. Desouvrey for pointing these facts out to me. 31. See also Koopman and Lefebvre (1982, 66): “ap apparait dans deux positions: MODE et ASPECT.” 32. Leblanc (1989, 42ff) notes that, in contrast to these adverbs, which can occur between INFL particles, neither place nor manner adverbs can (they rather appear in sentence final position). Michel DeGraff tells me that for him (37a—b) are almost impossible (completely, if the adverb intervening between the two aps is missing). In fact, variation concerning the sequence of two aps and that of other identical morphemes is reported to exist in Lefebvre (1996) and is attributed by her to a general filter holding in one but not the other variety. 33. Koopman and Lefebvre (1982, 65) list a under Mode (together with ap and pu), adding that “la différence entre a et ap reste 4 Etre investiguée.” Forrest (1990), and Lefebvre (1995, 160ff; 1996, §1.1) characterize the difference as one between “potential” or “irrealis” versus “definite” or “realis” future (viendra peutétre / éventuellement versus viendra), Spears (1989, 209) as one between subjunctive versus indicative future. Comparable distinctions are found in the Austronesian language Kusaiean (Frawley 1992, 361) and (as pointed out to me by Paola Beninca’ and Cecilia Poletto) in the Romance variety spoken in Inverso Pinasca (Piedmont, Italy). The latter also has an irrealis form of the past (Griset 1966, 70). ‘Irrealis’ future may be analyzed as a combination of Mood,,,.,,,, and T(Future) (with Mood,,,cans raising to T(Future)). The two are realized separately in certain languages, with T(Future) preceding Mood,,,..,3,;. See, for example, the cases of Ndyuka (another creole language of Surinam) in (1), and of Tigak (a Papuan language) in (11):

(1) [ben o sa poi (fu) nyan ete. (Huttar and Huttar 1994, 519) 2sg PAST FUT IRR be-able (for) eat yet. “Would you have been able to eat already?’

192 NOTES TO PAGES 62-64

(ii) a. Vo nak ima (amaua). (Beaumont 1979, 79) FUT Isg come (tomorrow) ‘T will come (tomorrow).’ b. Vo nak min ima (amaua). FUT 1|sg DUB come (tomorrow) . ‘I may come (tomorrow).’ In the light of (1), the future particle sa of Sranan (related to Ndyuka) can be literally taken to be generated in Mood,,,ea,5 and to raise to T(Future).

34. This appears to correspond in a mirror fashion to the order of suffixes in Inuit (Eskimo-Aleut). See example (ia). Mennecier et Tersis (1988, 202) explicitly note that the future tense suffix cannot precede the modal pouvoir (see (ib)) (we have translated and adapted the glosses):

(i) a. Nii-sinnaa-ssa-vu-q. eat-CAN-FUT-predicative-M.P.3. ‘I will be able to eat.’ b. *nii-ssa-sinnaa-vu-q. eat-FUT-CAN-predicative-M.P.3 35, This conclusion is shared by DeGraff (1993, 32ff). The deontic sense is also forced if dwe follows the negation pa, while the epistemic sense is forced if it precedes pa. MagloireHolly notes that the same holds for ka/kapab ‘can’ in interaction with te and pa. 36. See also the possibility in Italian of an (imperfect) past form of dovere (‘have to’) with an epistemic interpretation:

(i) Allora, dovevano essere (state) le cinque. ‘Then, it had to (have) be(en) 5 o’clock.’ (‘Then, it was probably 5 o’ clock’) Koopman and Lefebvre (1982) give a comparable example (m te dwe te vini), translating it once as ‘J’aurais dfi étre venu’ (p. 72) and once as ‘Je devais étre venu’ (p. 89). 37. Certain interpretive facts may suggest that the second (T(Anterior)) te (henceforth te») may get deleted, as was the case for Sranan ben,. For example, Magloire-Holly (1982) translates Zd te dwe maze in the same way as she translates Za te dwe te vini: ‘aurait dQ’. In addition, fe plus nonstative verbs is translated with a past perfect rather than with a simple past (as in Sranan again): e.g., Mwen te rakonte yon istwa ‘Thad told a story’ (= 13b of Leblanc 1989, 46). See also the discussion in Lefebvre 1996, §$1.1, 1.2, and references cited there). For a case where T(Past) and T(Anterior) are both systematically realized (with two different particles), see the case of Seychelles Creole in Appendix 2. 38. For Fongbe, see Avolonto (1992a,b), Kinyalolo (1992), Lefebvre (1995, 1996), and references cited there; for Gungbe, Aboh (1993, 1996), from which most of the examples in the text are taken (others having been directly provided by him in personal communications). 39. This aspect is often called “prospective” (see Comrie 1976a, 64ff.; Bliese 1981, 117ff; Davies 1981, 31; Dahl 1985, 111ff; Frawley 1992, 322; Kinyalolo 1992; Vonen 1994, 387; Lefebvre 1995): “Languages encode not only the beginning and end of an event, but also a point just prior to the beginning of an event. This is known as prospective aspect’ (Frawley 1992, 322). In the literature, other terms for the same aspect are proximatif, proximative (Cloarec 1972, 110ff; Heine 1994), impending (Abbott 1991, 120), unrealized (Robinson and Armagost 1990, 318), and immediate (future) (Mithun, forthcoming). Heine (1992, 339) referred to it as the ALMostT-aspect. Here, I follow the prevailing terminology, reserving “proximative” for the so-called soon-aspect (the converse of “retrospective,” or JUST-aspect). See chapter 4 for further discussion. Cram (1983/84) uses the term “prospective” to characterize the verbal aspectual particle gu (lit. ‘to’) of Scottish Gaelic. However, it is dubious that Gungbe na and Scottish Gaelic gu refer to the same kind of aspect. First, Cram sometimes renders it as ‘is to’ (see Tha Calum gu bhith faiceallach ‘Calum ts to be careful’), in what looks like a modal meaning. Second,

NOTES TO PAGES 64-67 193

gu precedes the progressive particle, whereas nd follows the progressive particle. Its ability to either precede or follow the perfect particle (Ramchand 1995) in fact contrasts with the rigid order of the other aspectual particles and may indicate that it is a main predicate. As usual, particular care must be taken concerning the varied terminology for aspects employed in the literature. 40. Avolonto (1992b, chap. 5) postulates an “injunctive projection” for (Fongbe) ni (and other modal particles) “qui encodent dans leur sémantique le sense d’un ordre” (p. 53). See also Avolonto (1992c). Aboh (1996) proposes to analyze (some uses of) nf as a complementizer. Aboh (1993) recognized an additional particle in Gungbe, marking what he called “aspect révolue” (‘temporal priority’): kd. More recently, he has come to favor an analysis of ko as an AdvP, which he renders as ‘already’ (see Aboh 1996; see also Avolonto 1992b, 36ff and Lefebvre 1995 on the adverbial nature of ké in Fongbe):

(i) Séna ko x3 méto 1S. S. already/temporal priority buy car the ‘S. has already bought the car.’ In addition to habitual, progressive, and prospective aspect, Aboh (1993, 1996) discusses another aspect (perfect aspect), which is realized as zero in Gungbe. Example (11), for instance, is interpreted as perfect even in the absence of any overt particle:

Gi) Séna x5 méto 15.

S. buy car the ‘S. has bought the car.’ 41. I take the auxiliary (bod ‘be’) in between the two particles to be inserted under an intermediate functional head. Perhaps wedi selects a nonfinite verbal form, a requirement that

the particle/preposition yn cannot satisfy. One could possibly view yn not as the head of ASP prooressive but as the head of the CP/PP above it (in Kayne’s 1993b analysis); a possibility

indirectly supported by the Italian (and Spanish) progressive periphrasis stare ‘stay’ + V-ndo, if analyzed as in (1): (i) lL BE [ CP/PP Proc [ progressive Aspect -ndo [ so [ Vv ] | | J ] (with Py oct BE—> stare

‘stay’)

42. For the similar case of Irish (ia) and Scottish Gaelic (ib), see O Baoill (1994, 209) and Cram (1983/84, 315), respectively:

(i) a. Bhi méi ndiaidh a bheith ag scriobh leabhair. BE-PAST I after to be at writing book-GEN. ‘I had Gust) been writing a book.’ b. Tha mi air a bhith a’ sgriobhadh litir. am I PERF (lit. ‘after’) to be PROG (lit. ‘at’) write a letter. ‘[ have been writing a letter.’

The Celtic order thus matches that of English and Spanish seen in (20) and also the order found in the Niger-Congo language Temne. Cf. (ii) from Kamarah (1997, 33): (ii) 13 po yird ko-ko I FUT PERF PROG go ‘I will have been going.’ 43. See also (i), which Yavas (1980, 63) says is slightly marginal “because there is a less cumbersome way of saying the same thing”: G) ??John diin calis-ryor ol-mug ol-mali J. yesterday work-PROG be-PERF/ANT be-must. ‘John must have been working yesterday.’

194 NOTES TO PAGES 67-70

In recent work, Kornfilt (see Kornfilt, 1996) argues that most agglutination in Turkish is in fact apparent, arising from the juxtaposition of constituents formed by a verb with a closing suffix followed by (sometimes null) clitic auxiliaries with other closing affixes. To judge from

certain common properties, it is possible that her arguments carry over to Korean (and to agelutinating languages, more generally). See also Dobrovoisky (1976). 44. “Conjunct” prefixes are often opposed to “disjunct” prefixes, which precede them, and are separated from them by a stronger phonological boundary (see Rice 1991a,b, and references cited there), On a speculative remark on “disjunct” prefixes, see text at (68). 45. The problematic nature of the Navajo case for the Mirror Principle is noted, and discussed, by Speas (1990a,b). In a nonrestrictive theory that allowed both lowering and right- (as well as left-) adjunction, the Navajo case could be derived by lowering T and adjoining it to the right of Asp, and then lowering the Asp-T complex head, adjoining it to the left of V. Needless to say, in such a system almost anything goes, including many unattested cases. Here, I am assuming, after Kayne (1994), that only (single) adjunctions to the left are possible, of heads (to heads), and XPs (to XPs), subject to Relativized Minimality (Shortest Movement), including the successive leftward adjunctions of portions of clausal, and nominal, structure in “headfinal” languages. See the cases discussed in §§ 3.1 and 3.2 and in Cinque (1996, §6). The latter adjunctions cannot be to adverb-related functional XPs (to which AdvPs are adjoined). So they must be adjunctions to separate XPs, whose nature remains to be clarified. 46. Possible evidence for this conjecture may be the fact (discussed in Rice 1991a) that the “prefixes” in (67), traditionally called “conjunct” prefixes, are a domain of (phonological) rule application that excludes the verb (and the so-called disjunct prefixes preceding them).

47. See also the detailed studies by Rice (1989) on Slave and by Axelrod (1993) on Tense, Mood, and Aspect in Koyukon Athapaskan. 48. I am grateful to Peter Culicover for a clarifying discussion. A comparable case is represented by Yimas, a polysynthetic language of New Guinea. In Foley’s (1991, §6.4) description, the verb root, which is followed by Aspect, Tense, and Agreement suffixes (in that order), is preceded by the following sequence of (optional) prefixes: modality-negationobject pronominal-subject pronominal-definitive-incorporated adverbs-directionals-valencyincorporated object. Some adverbs incorporate optionally (see (ia—b)); others obligatorily (see (iia—b)), where

“obligatorily incorporated adverbials commonly, but not exclusively, express aspectual or temporal notions, while optionally incorporated adverbials denote the manner or degree to which an action is carried out. [In example (iib),] the adverbial nanay— DUR 1s... . in combination with the aspectual suffix -na(y) DUR.” (Foley 1991, 336f):

(i) a. Na-n-mampi-ira-wampupkra-ntut. 3sg-O3sg-again-ALL-angry-(Remote)PAST. ‘He was angry with her again.’ b. Mampi na-n-ira-wampunkra-ntut. again 3sg-O3sg-ALL-angry-(Remote)PAST. ‘He was angry with her again.’

Gi) a. Impa-yakal-cmi-kiantut. 3DUAL-CONT-say-(Far)PAST. ‘Those two were talking.’ b. Kay na-nanay-yamat-pfia-ntut.

canoe 3sg-DUR-carve-DUR-(Remote)PAST. ‘He was making a canoe.’ 49. For the Bantu language Shona (Bellusci 1991), the null auxiliary analysis of the Navajo conjunct prefixes might be more relevant, for aspect prefixes precede tense prefixes, as in Athapaskan:

(i) Nd-a-ka-swero-funda. (Bellusci 1991,29) AGR-ASP-PAST-all day long-study. ‘T studied ali day jong.’

NOTES TO PAGES 70-73 195

50. In keeping with the fact that habitual aspect is cross-linguistically the “highest” of all aspects, Bybee notes that it does not appear to be expressed derivationally. Rather, “it is either inflectional or marked with free grammatical morphemes (such as auxiliaries)” (p. 101). But see the case of Aleut in Appendix 2, where, according to Bergsland, all aspects are expressed derivationally (including habitual aspect; see example (16a)), while tense is inflectional. 51. It also appears that functional notions encoded via modifications of the verb root (as in Semitic nonconcatenative morphology) correspond to functional heads “lower” than those encoded via affixation (see Bybee 1985, 34). So, for example, in Arabic, Voice and Aspect is nonconcatenative while mood is affixal (Fassi Fehri 1993, 82ff). Similarly, in Chaha (a South Ethiopian Semitic language) aspect is nonconcatenative while tense in affixal (Rose, 1996). This follows naturally if features within the root have to be checked “before” features on outer morphological layers. What remains to be ascertained is to what height in the hierarchy functional notions can be coded as modifications of the root. In the same vein, it is to be expected that functional notions encoded on the lexical verb (either through root modifications or affixation) systematically correspond to functional heads “lower” than those expressed via independent particles or auxiliary verbs. 52. Evaluative mood suffixes and particles exist in other languages as well. See chapter 4 (84.5). 53. Palmer (1986, 65) says that “the ‘non-past uncertain’ form of the verb appears to be used as a general marker of epistemic modality.” I assume that the evidential head has been crossed over by its complement (see Kayne’s 1994 analysis of “head-final” languages). 54, Recall that the apparent evidence from Haitian and other Creoles for the same order is perhaps irrelevant (see §3.3). Hidatsa (North America) provides evidence for Mood, gentat > T(Past) (expected by transitivity if Mood, gentiat > MOdepisteme ANd MOdesteme > T(Past).

See (39a) in Appendix 2). Una and Tauya (New Guinea), instead, provide evidence for Modensteme > T(Future), which is also expected by transitivity if Mod. stemic > T(Past) and T(Past) > T(Future), for which see (76)-(81).

55. If the past auxiliary is followed by the verb in the unmarked form, it yields past progressive (see (ia)); if it is followed by the verb in the anterior (= past) form, it yields the anterior of the past (see (ib)), all from (Louwrens, Kosch, and Kontzé 1995, 52ff):

(4) a. Mosadi o-bé a-réka nama. woman AGRs-AuxPAST AGRs-buy meat. ‘The woman was buying the meat.’ b. Mosadi o-bé a-rék-ilé nama. woman AGRs-AuxPAST AGRs-buy-ANT meat. “The woman had bought meat.’ As expected, the anterior of the future is rendered as in (11), while (iii) is the future progressive. Present progressive is marked with another prefix. See (iv):

Gi) Mosadi o-tla-ba a-rék-ilé nama. woman AGRs-FUT-Aux AGRs-buy-ANT meat. ‘The woman will have bought meat.’

(iii) Mosadi o-tla-ba a-réka nama. woman AGRs-FUT-Aux AGRs-buy meat. ‘The woman will be buying meat.’

Gv) Mosadi o-sa-réka nama. woman AGRs-PROG-buy meat. ‘The woman is still buying meat.’

196 NOTES TO PAGES 73-75

Note that the translation of example (81) would lead one to expect a-tla-rék-ilé (FUT ANT) rather than a-tla-réka (FUT). 56. The Australian language Bagandji has a suffix of “uncertain possibility” preceding the future tense suffix (Hercus 1982, 197):

(4) Bada-nda-d-uru-ana. bite- UNCERT POSSIB-FUT-3sgtr-3sgobj. ‘It might bite him.’ Also see the case of Tigak cited in n. 33. The Jacaltec example cited in Palmer (1986, 45), may be relevant as well:

(ii) x'-'oc heb 1x say-a' hun-wj munlabal. ASP-start PL woman look for-F UT a-IRR pot. ‘The women started looking for a pot.’ 57. The claim that cde is irrealis mood, rather than future tense, is Svantesson’s (1994, 276). To judge from Rottet (1992, n. 12) (irrealis) mood also precedes root modals in Lousiana Creole. Hawick Scots gives evidence for the order T(Future) > Mod,,,, (expected by transitivity if T(Future) > Mood,,,caj,5 and Mood, peas > Mod root):

(i) He’ll can get you one (Brown 1992, 75) (= He will be able to get you one) 58. For the frequent inceptive interpretation of perfective forms, see Comrie (1976a, 19ff). 59. See also Thurgood and Thurgood (1996). A clue to distinguish T(Anterior) from perfect aspect may also be given by the translation, when the relevant morphology co-occurs with (absolute) past tense morphology. If it is translated as simple past (as in Dagaare [see (1)]; SanioHiowe [see example (59b)], repeated here as (11); or Basque [see (iii)], it is presumably perfect aspect, not T(Antertor). If it is translated as “pluperfect” (anterior of the past), it is presumably T(Anterior), though T(Anterior) morphology sometimes expresses perfect aspect syncretically.

G) Oda kul-ee (versus kul-o) la. (Bodomo 1993, 32) (s)he PAST go-home-PERF (versus go-home-PROG) affirmative marker ‘(S)he went (versus was going) home.’

(ii) Onu tei ye. (Lewis 1972, 17)

sit PERF PAST “We sat down.’ (ii) Nik liburnak irakurr-i n-it-u-en. I-ERG books-ABS read-PERF isgE-PL-have-PAST. ‘I read the books.’ 60. Malay sudah, however, may be the adverb “already” rather than a T(anterior) particle. This is suggested by the fact, noted in Sneddon (1996, 202), that sudah harus and harus sudah both mean ‘must already’ (see (ia—b)); which can be taken to involve raising of the modal past the adverb:

(i) a. Kamu sudah harus di sini jam lima. b. Kamu harus sudah di sini jam lima. “You must be here by ftve o’ clock.’ 61. Guyanese Creole shows the order AsPaurative > ASPprogressiver Where durative aspect “refers to the fact that a given situation is CONCEIVED OF as lasting for a certain period of time” (Gibson 1992, 57). Since durative aspect follows habitual aspect and Guyanese has no particle for perfect and retrospective aspect (I assume don to be T(Anterior)), the exact location of durative aspect with regard to the former aspects 1s underdetermined. For concreteness, I will order it between retrospective and progressive aspect. Some clue will come from the order of the corresponding adverb classes, to be discussed in the next chapter.

NOTES TO PAGES 75-78 197 The same relative order between Aspgurative Ad ASP progressive: once the order of suffixes is

“undone,” is attested in the Papuan language Nankina, where “usually the durative aspect is used to convey the idea that the situation lasted for a long period of time” (Spaulding and Spaulding 1994, 60). Cf. (I): GQ) Nin wo Kap jipm-ripm-gwain-yn-Ku yaka pA-nam. Ip go.up possum kill-PROG-DUR-same subj-DefSEQ(uential aspect) again come.down-INDEFIp ‘We will go up and be killing possums for a while then will come down again.’

62. Evidence to the same effect is apparently found in Comanche. Robinson and Armagost (1990, 318) translate the suffixes -turih and -wa ?ih of ‘unrealized’ aspect with be about to or almost (see (ia—b)), crucially saying that “both suffixes can take a following -t# (progressive) to express ongoing potentiality, and both are often translatable as the English future” (p. 318). See (ii).

(i) a. ... puhihwu-hima-ruri-ka. money-take-UNR[PROSP]-DS. *,.. were about to receive money.’ b. Noha u? nv kwuhe-ruri. nearly D4NS 1xAs catch- UNR[PROSP] ‘He almost caught me.’

(ii) ...ke tamu-ma-tu musa-sua-war-tu. NEG 1IND-on-along ?-think-UNR[PROSP]-PROG. ... will not be concerned about us.’ Also see the case of the West African language Akuapem in (iii), from Campbell (1996, 90) where the prefix following the progressive prefix is called “future,” but rendered as “about to’:

(iii) Kofi a-tO bayerE re-bE-di. K. PERF-buy yam PROG-FUT[PROSP]-eat. ‘Kofi has bought a yam and is about to eat it.’

63. The position of passive Voice in Latin, a suffix (-r), external to (future and past) tense and agreement, would seem to represent a counterexample to the low location of Voice in (96). This suffix, however, has middle and impersonal usages similar to the impersonal se/ si of Romance, which may support an analysis of it as a higher impersonal (passive) clitic Voice. See Calabrese (1985) for a diachronic argument to this effect. (I am indebted to Massimo Vai and Andrea Calabrese for bringing this problem to my attention.) 64. A further dilemma is whether the hierarchy is always present, in all sentences of a language, or only those portions of the hierarchy are present which are needed to accomodate morphological material (affixes, adverbs, etc.). Once again, I will try to argue for the second, stronger, alternative, on the basis of the most natural interpretation of the marked and default values of each head (see chapter 6).

Chapter 4 I thank Paola Beninca’, Gerhard Brugger, Peter Culicover, Hubert Haider, Richard Kayne, Cecilia Poletto, and Sten Vikner for comments on different parts of this chapter. 1. The notions of “alethic” necessity and possibility used here include “contingent” (it is necessary that you hand in your homework by tomorrow, It is possible that it will not rain) as well as “logical,” necessity and possibility. What latridou (1990) calls “metaphysical” modality seems to bear some resemblance to the logicians’ notion of alethic modality.

198 NOTES TO PAGES 79-81

2. In Turkish, where the corresponding modal notions are expressed by suffixes, one finds the exact mirror image of Scots. See (ia), provided by Jaklin Kornfilt, who also pointed out that for her the future suffix is perhaps somewhat less natural than the aorist suffix (see also (ib), from Yavas 1980, 160, and Kornfilt 1997, 375):

(i) a. Gel-e-me-yebil-ecek. come-ABIL-NEG-POSSIB-FUT. ‘(S)he may in the future not be able to come.’ b. ?Mary-nin Paris-ten al-mis ol-abil-eceg-i elbise cok giizel. M-GEN P-ABL buy-PERF be-POSSIB-FUT-3ps dress very pretty ‘The dress which Mary may have bought in Paris is very pretty.’ 3. “Will does not co-occur with epistemic must or would’ (Brown 1992, 78). The modal should, however, can (which raises the question whether in such cases should is a modal of alethic necessity rather than an epistemic modal). Additional evidence that epistemic modality is higher than alethic modality is possibly given by their position with respect to negation. While negation can precede alethic modality (it can’t be five o’clock: NEG > POSSIBLE, with incorporation of can to n’f), negation necessarily follows epistemic modality. See chapter 5. Another difference between the two 1s the well-formedness of alethic modals and the 1li-formedness of epistemic modals in interrogative sentences (Could/Might Guido know the answer? versus * ?Must Guido know the answer?; from Spellmire 1994). See also §4.7 and §4.10. 4. Here, dovere can also be interpreted epistemically, with potere retaining a pure possibility (alethic) interpretation: Questo dovrebbe potersi verificare presto ‘It should be the case that it is possible that this happens soon’. This judgment differs from Picallo’s (1990, 294ff) on the sequence deu poder in Catalan. 5. I abstract from the further, purely epistemic, interpretation of dovere in (7b) (‘G. probably wants ...’). The ungrammaticality of the Danish example is perhaps due to ma allowing for a deontic interpretation, but not for one of alethic necessity. Dovere preceding volere in Italian may also have the particular interpretation of “conjectural necessity” which Beninca’ and Poletto (1994, 1996) recognize for such cases as (ib), with Disogna (‘it is necessary’).

(1) a. Deve proprio volerci aiutare, se fa cosi. ‘He must really want to help us if he acts like that.’ b. Bisogna proprio che voglia aiutarci se fa cosi. ‘It is necessary (it is really the case) that he wants to help us, if he acts like that.’

This usage is not epistemic, as they note, since it is possible with bisogna, which does not allow standard epistemic usages (*Bisogna che siano le cinque ‘It must be five o’clock’). See Beninca’ and Poletto (1994, n. 9). 6. A similar conclusion (modulo the term “epistemic possibility” for what is here called “alethic possibility”) is reached in Nuyts (1993) on the basis of the Dutch sentences in (a):

Gi) a. Jan kan morgen weg moeten. ‘J. may have to go tomorrow.’ b. Jan moet morgen weg kunnen. ‘J. must be able to go tomorrow.’ “[E]pistemic modality should clearly be higher in the hierarchy than deontic modality (and facultative modality should be even lower in the hierarchy)’ (p. 961). 7. Whitley (1975) observes that in the may can sequence found in Southern American varieties “may can be interpreted only in the sense of possibility, never permission” (p. 105). 8. Its most serious conceptual difficulty is its unconstrained character, which fails to predict the existence of all and only the tenses actually found cross-linguistically. The absence of a past past past of the future, for example, does not follow from any internal property of the system. Other conceptual and factual defects are discussed in the references mentioned in the text.

NOTES TO PAGES 81-83 199

9, The essence of Reichenbach’s system is the following. Nine different tenses are derived from the different relations which three primitive times (the “time of speech,” S; the “time of the event,” E; and a “reference time,” R) enter on the time axis: whether one pre-

cedes (E R), follows (R E), or coincides (E,R) with the other(s). The different tenses are given in the following table, adapted from Vikner’s (1985, 82):

Reichenbach’s names Traditional names Representations Examples

1. simple present present E.R,S works

2. simple past past/preterit E,R_S worked

3. anterior present present perfect E_R,S has worked

4. simple future future S_R,E will work

5. posterior present future S,R_E will work 6. anterior past past perfect ERS had worked 7. posterior future —. S RE will be going to work 8. posterior past future of the past RES would work

(R_E,S) "" ft" (R_S_E) 9. anterior future future perfect ESR will have worked (S,E_R) (S_E_R) "" "" This yields a total of nine (supposedly universal) tenses (technically 13, if the multiple representations for the future perfect and the future of the past are taken into consideration). 10. Reichenbach exemplifies it with the Latin future of the “future participle” in -urus (abiturus ero ‘I shall be one of those who will leave’), saying that it “[speaks] not directly of the event E, but of the act of preparation for it” (p. 297): an aspectual property, more than a relation between times. See Jespersen’s (1924, 263) critical remarks on Madvig’s similar proposal for such a tense in Latin and the discussion on the so-called prospective aspect in n. 39 of chapter 3, and below. 11. T(Anterior) corresponds to Comrie’s (1981, 25ff) “relative past tense” (namely, relative to a point of reference distinct from “now’’), and T(Past) to “absolute past tense” (which is relative to the point of reference “now’’).

12. Notice that the time of the event (“the receiving of the paper,” in Vikner’s example in the text above) is not ordered with respect to R, ((November’) or S (the speech time); nor is R, (‘by the first day of term’) ordered with respect to the speech time. This is an apparently correct result. See Vikner (1985) for discussion. The “future (perfect) of the past” uses of the “conditional” are characteristically found in narrations (“John left for the front. He would never return”) or in “sequence of tense” contexts in indirect speech (“He said that he would (have) come’). See Comrie (1981, 27). On “sequence of tense’’ phenomena, which remain outside of the present discussion, see Eng (1987), Hornstein (1990, chaps. 4, 5), Binnick (1991, SOff, 35311), and Stowell (1995) and references cited by them. 13. This system indeed holds in Danish, Dutch, French, German, Italian, Hawaiian, Yoruba (Vikner 1985, 84), and Bulgarian (to judge from Lindstedt 1985). The distinction between the “imperfect” and the “simple past” found in Romance (and Bulgarian) seems to be aspectual rather than temporal. See Vikner (1986). 14. So, for example, one should explain why the universal system is not (1) (with a T(Anterior) embedded under another T( Anterior) embedded under a T(Future)), or some other arrangement of the three relations, instead of (1):

200 NOTES TO PAGES 83-84

Gi) R, S; R, R,; E R, qi) S R,; R, R,; E R, By letting their (unique) R coincide, precede, or follow S and E, Giorgi and Pianesi (1997) allow for the nine basic tenses of Reichenbach (1947). While avoiding the problem connected with the adoption of one three-place relation, their system is open to essentially the same criticism moved to Reichenbach’s by Vikner (1985). For example, it allows for two distinct representations of the future (S__§ _R+R,E, and $S,R + R___E) and for a future of the future (S__R-+R___E), while making no room for the future perfect of the past (would have received). Comparable remarks hold for Hornstein’s (1990) system, which, in fact, allows for an even greater number of tenses by treating the two possible linearizations of the times associated by a comma as two distinct representations (e.g. E,R # R,E). Here I will not consider how Vikner’s theory relates to Stowell’s (1996) (see also Zagona’s 1995) treatment of tenses as predicates ordering two temporal points (e.g., past = (roughly) S after E), except to note that Vikner’s system appears to be easily translatable into Stowell’s. 15. Some modifications and additions to Vendler’s original classification are proposed in the literature, but I will ignore them here, as our focus will be on the other kind of aspect. See Dowty (1979), Freed (1979), Mourelatos (1981), and Smith (1991), among others. “Accomplishments” and “achievements” differ from both “activities” and “states” in having an inherent natural end point (say, when the mile 1s finished, or the top of the mountain is reached). As such, they are bounded (telic), while “activities” and “states” are inherently unbounded (atelic). Even if the latter too have a beginning or an end, that is not part of their meaning. “Accomplishments” differ from “achievements” in that they are made up of internal stages, which represent advances toward the final point (Smith 1991, 49ff). “Achievements,” instead, are instantaneous changes of state (only having preparatory stages, if any). “Activities,” differently from “states” (which present a situation as not involving stages or changes) are made up of stages (of the same nature as the whole). A well-known test distinguishing “activities” from both “accomplishments” and “achievements” is the compatibility of the former, and incompatibility of the latter, with “durative adverbials” introduced by “for” (and vice versa for those introduced by “in’’). See the references cited in the discussion here and §4.20 for more discussion on durative adverbials. 16. The grammatical means may be affixes (typically on the verb), particles, auxiliaries, stem modifications, and others. 17. This is so because to belong to a certain class (say, of “accomplishments” or “activities”) depends not only on the lexical meaning of the verb but also on the internal make up of its arguments, as pointed out by Verkuy! (1972). For example, a bare object DP, as opposed to a definite one, renders the event an “activity” rather than an “accomplishment”: He ate apples for (*in) a month versus He ate the apples (??for) in a month. For recent discussion on these questions, see Tenny (1987), Borer (1996). 18. The grammatical speech-act moods are only indirectly related to the speech (or illocutionary) acts themselves. For example, a sentence in the interrogative mood can, be used to express a command; one in the declarative mood can be used to make a promise. Here, I will ignore the complex relations between linguistic form and the illocutionary force of the sentence. 19. This, in fact, remains to be seen, Hengeveld (forthcoming b, §2.2) mentions the existence of particles that serve to modify (emphasize or mitigate, often with politeness implications) the illocutionary mood suffixes. See also the Korean peripheral suffix -yo of example (3) in chapter 3, and Bybee’s (1985, 185) observation that in most cases an emphatic affix “occurs as one of the first prefixes (as in Kutenai, Pawnee and Kiwai) or as one of the last suffixes (as in Dieguefnio, Tarascan and Kwakiutl).”

NOTES TO PAGES 84-87 201

20. Other such “duplications” between the IP and CP “spaces” seem to exist. In certain languages (e.g. Irish; see McCloskey 1996), tense distinctions appear not only on verbs but also on complementizers. This, along with other reasons, is the basis for Rizzi’s (1997) postulation of a FiniteP in the complementizer “space,” coexisting with a TP in the IP “space.” Other languages (e.g., Basque; see Laka 1990) show negative complementizers in addition to negation in the IP “space.” 21. See also the so-called admirative mood of Albanian and other languages, which expresses surprise for an unexpected event (Kallulli 1995; Friedman 1986, 180ff; Delancey 1997). Derbyshire (1986, 527) reports the existence of an evaluative particle (kabay ‘it is good that’) in Brazilian Arawakan. Thurgood (1986, 216 ff) mentions the existence in the TibetoBurman language Akha of sentential particles expressing such contrasts as “surprise/nonsurprise,’ “luckily/unfortunately.’ An affix (-ni), meaning that the speaker considers the content of the proposition “unfortunate,” is found in the Amerindian language Piro (see Wise 1986, 587). In discussing the two Chinese future auxiliaries hii (lit. ‘can’) and yao (lit. “want’), Myhill (1992) cites text-based data “showing that hai is particularly associated with positive evaluations, as 41 of 100 clauses with hai could idiomatically add xingkui ‘fortunately’, while only five of 100 clauses with yao could” (p. 86). 22. The term “evidential” seems to have been first used by Jakobson (1957/1971) in a presentation of Boas’s work on the subject (see Jacobsen 1986, 5). 23. In this context, it is tempting to take the verb in the conditional mood, which expresses the “future of the past,” to raise to the evidential mood head, to check the relevant feature (recall the evidence from chapter 3 that the evidential mood head is higher than the T(Past) and T(Future) heads). Under this perspective, in fact, such a usage of the “conditional” would not be possible if the evidential mood head were lower than T(Future) and T(Past), as lowering is not admitted. More generally, it is to be expected that possible acquisitions of new functional values by a head (synchronically or diachronically) will be a function of the universal hierarchy of heads, and of the limitation to upward movements only. 24. [ignore here the (quite) different “evidential’’ nuances of these (and other) adverbs. For an in-depth study of the complex conditions of use of one such adverb (obviously), see Michell (1975). 25. Because of that, various authors refer to root modality as “subject- or agent-oriented”’ modality and to epistemic modality as “speaker-oriented” modality (see Bybee 1985, 166, Heine 1995, among others). Although certain authors include evidential knowledge in the knowledge relevant for the speaker’s epistemic judgments (see, for example, Palmer 1986, 51 ff), [have kept the two separate here, for reasons already seen in chapters 1 and 3 (the two categories can co-occur). Nuyts (1993, §1.3) also argues for keeping evidential and epistemic notions separate (with the second in the scope of the first). I ignore here the possible more abstract common logical structure unifying the epistemic and root uses of modals. For discussion, see Kratzer (1977, 1981). 26. Even adverbs like certainly and surely, while expressing an even stronger conviction on the part of the speaker, still betray a lack of knowledge. See John is certainly home, now as opposed to the unmarked John is home, now. Only the latter presents the proposition as a fact. See Lyons (1977, 809), Perkins (1983, 23 and n. 3, with relative text), and Bybee (1985, 180). 27, See Comrie (1985, 51) and Aboh (1996, 23), among others. Comrie also notes that in Mam and Jamaican Creole “it is usual to omit tense markers when an overt adverbial of time location is present” (p. 31). Tense distinctions may also be surrogated by aspectual distinctions (perfect/imperfect). See Comrie (1976a, 82ff) and Binnick (1991, 434 ff). 28. Some languages show quite elaborate “metrical” (or “remoteness’’) tense systems. Dschang, a Bantu language of Cameroon, distinguishes five pasts and five futures (i.e., five

202 NOTES TO PAGES 87-88

different, and symmetrical, time intervals in the past and in the future). See (1) and (ii), from Chia (1976, 68f); see also Hyman (1980) and Comrie (1985, 97):

(i) a. OOkonS més. you PAST, like child “You have just liked the child (a few minutes or hours ago).’ b. O OO n‘konds més.

you PAST, like child “You liked the child (sometimes this morning).’ c. O ke kond mé. you PAST; like child “You liked the child yesterday.’ d. O le kond mé. you PAST, like child “You liked the child the day before yesterday (last week, etc.).’

e. O leld n’kond mS.

you PAST; like child “You liked the child years ago.’

Gi) a. O ‘Okons més. you FUT, like child “You will like the child in a few minutes or hours.’ b. O pin kond més.

you FUT, like child “You will like the child this afternoon or tonight.’ c. O lu kons md. you FUT; like child “You will like the child tomorrow.’ d. O ‘dé? kond mé. you FUT, like child “You will like the child the day after tomorrow.’ e. O ‘fi kond mS. you FUT; like child “You will like the child in years to come.’ Similar systems are attested in North American languages (see Mithun forthcoming), and other language groups (see Comrie 1985, chap. 4 and Frawley 1992, 363, and references cited there). I assume that such distinctions are orthogonal to the only “real’’ Tense distinctions discussed earlier, which involve different relations among the E,R,,R,, and S points. 29, If “bare” adverbial DPs are headless PPs rather than adverbs (see, however, Riemsdijk 1996, n. 11 and p. 18), the fact that they cannot appear in Spec,T(Past) or T(Future) may be no real problem, as PPs are generally barred from the IP functional “space” (see Jackendoff 1977, 73). If they are “predicates,” as argued in Stroik (1992)——predicated of the verb’s empty

temporal argument—their nonoccurrence within the IP functional “space” is likewise not surprising, as other (secondary) predicates also seem excluded from there. Their distribution is similarly unsurprising if they pattern like argument DPs, as Giorgi and Pianesi (1997, chap. 3,

§2.5) suggest. For general discussion on their distribution and categorial status, see Larson (1985), McCawley (1988), Stroik (1992), Giorgi and Pianesi (1997), and Nilsen (1998, §2.61). 30. According to Botne’s (1993) glosses, the Bantu language Tumbuka appears to have a Present lower than T(Future) (see (1)). His description makes it clear that the form encodes the default value of T(Anterior): “The lexical verb that appears as the complement of -va may be marked as either simultaneous with (-ku-) or anterior to (-a-) the second tense locus referred to by -va” (p. 12):

(i) a. Va-zamu-va va-ku-vina. 3pl-FUT-be 3pl-PRES-dance ‘They will be dancing.’ b. yi-zamu-va y-a-baba.

Agr-FUT-be agr-ANT-give birth ‘will have given birth.’ The higher position which ora (and adesso ‘now’) can occupy, before all sentence adverbs (including speech act ones) was taken in chapter | to be the position of “adverbs of

NOTES TO PAGES 88-90 203

setting.” Ora/adesso, however, do not sound impossible between evaluative and evidential adverbs (?Gianni fortunatamente ora chiaramente potra aiutarci di piu ‘G. luckily now clearly will be able to help us more’). This may indicate that more structure is involved. Some languages, in fact, show present tense morphology co-occurring with, and higher than, T(Past) morphology. See the cases of Aleut and Malayalam ((16), (18), and n. 7) of Appendix 2. Conceivably, T(Past) could be a relation between two reference points (R;__Ro), with T(Present) providing the anchor to S (Rp = S), when no sequence of tense is involved. The option Ry = E of the superordinate clause might account for the so-called subordinate present of certain languages (e.g., Tigrinya; Ké6s-Dienes 1984, 107), where a special present form is used in clauses dependent from a past verb to convey simultaneity; however, this is highly speculative, needless to say. 31. As opposed to purely epistemic modals and epistemic adverbs (see (ia-b)), alethic modals and alethic adverbs can occur in questions (see n. 3, (11) and Gi) here, and Schreiber 1971, 88; Corum 1974, 91ff; Doherty 1987; and Perkins 1983, 91ff), though Bellert (1977) reports an ungrammatical question with possibly:

(i) a. *Must it have rained? b. *Is it probably five o’ clock?

(ii) a. Could Guido know the answer? b. Mustn’t he know the answer?

(1) a. Aren’t Roman Catholic priests necessarily unmatried? b. Can he possibly have said that? 32. The property of “passive sensitivity” is more complex, however. Adverbs seem to behave (at least in part) differently from root modals. When the adjunct 9-role is incompatible with the subject, (pre-auxiliary) subject-oriented adverbs give no good output, apparently (*The flowers intentionally have been picked by visitors). Root modals, instead, appear to be able to assign the adjunct 6-role to the agent (Flowers may be picked by visitors) (see Jackendoff 1972, 105; Travis 1988, 305ff). Notice, however, that even subject-oriented adverbs can assign their adjunct 0-role to the agent, if they appear after the auxiliary (Flowers have intentionally been picked by the visitors). I will leave these problems to the side here, also recalling the observation in chapter 1, after Sueur (1978), that in French (and in Italian (Romance?)) the relevant adverb seem to be only agent-oriented. On the subject/agentorientation of adverbs and modals (which still needs to be completely understood), I refer for discussion to Jackendoff (1972), McConnel-Ginet (1982), Roberts (1985a,b, 1986), Travis (1988), and McCloskey (1996b, n. 8), among others. 33. Plausible candidates for the volition, obligation, and permission/ability classes are given in (1)-(i11), respectively:

(i) Intenzionalmente ‘intentionally’, deliberatamente ‘deliberately’ , (in)volontariamente ‘(in)voluntarily’, accidentalmente ‘accidentally’, etc. (ii) obbligatoriamente ‘obligatorily’, per forza ‘per force’, inevitabilmente ‘inevitably’, coercitivamente ‘coercively’, etc. (iui) legittimamente ‘legitimately’, col suo/mio/... permesso ‘with his/my/ . . . permission (permissibly), abilmente ‘cleverly/skilfully’, competentemente ‘competently’, goffamente ‘clumsily’, stupidamente “stupidly, etc. As different kinds of ability predicated of the subject, manner adverbs used in the “subjectoriented” way may fall into the class 111. 34. The volitional/nonvolitional affixes found in many languages (for example in the Uto-Aztecan language Cupefio | Hill 1969], in Lhasa Tibetan [Delancey 1986], in Tsova-Tush

204 NOTES TO PAGES 90-92

{Holisky 1987], in Sinhala [Inman 1992], in Malay [Wee 1995]) would seem to find their place in the head of ModP, jino,, although more work (e.g., concerning the relative position of the volitional affix) is needed before any conclusion can be reached. 35. As Comrie notes (1976a, 27) “a situation can be referred to by a habitual form without there being any iterativity at all” (see The temple of Diana used to stand at Ephesus). In (some) situations, habituality and iterativity are even incompatible (The lecturer stood up, coughed/*used to cough five times, and said .. .). See also Delfitto and Bertinetto (1994, 141, n. 6). 36. The same observation is made for French in Schlyter (1977, 92ff), where the following contrast is given:

(1) Habituellement Normalement ils regardent fréquemment la télé. Généralement ‘Usually they frequently watch TV.’

(11) Fréquemment * « Quotidiennement ils regardent habituellement la télé. Rarement ‘Frequently they usually watch TV.’

Things appear to be more complex, though. Richard Kayne (personal communication) tells me that for him both the orders often already and already often, as well as the orders usually no longer and no longer usually, usually still, still usually are possible. See (i1i)—(v):

(iii) a. I often have already seen it. b. I already often find myself accepting such offers.

(iv) a. ?They usually no longer win. b. They no longer usually drink much beer.

(v) a. They usually still have a lot of work left at the end of the day. b. They still usually drink a beer before lunch. Although di solito/solitamente ‘usually’ cannot (for me) follow spesso ‘often’, pia ‘no longer’ and ancora ‘still’, the apparently synonymous abitualmente can. See for example (vi):

(vi) a. Gianni (di solito) non prende pit (*di solito) la metropolitana. ‘G. (usually) no longer (usually) takes the subway.’ b. Gianni (??abitualmente) non prende pit (abitualmente) la metropolitana.

This suggests that habitual adverbs, like the other quantificational adverbs to be discussed next, can be “generated” in two distinct positions—a higher one quantifying over the event, and a lower one quantifying over the process. Italian seems to differ from English in having specialized adverbs for the higher (di solito/solitamente) and lower (abitualmente) scope positions. See n. 12 and 89 of chapter 1 and in the text for further discussion of these two quantificational adverb “zones.” 37. Habitual sentences must also be kept distinct from generic sentences. See the discussion in §4.21. Similarly, habitual adverbs must be kept distinct from other “adverbs of quantification” (often/rarely, always/never, etc.), which seem to correspond to distinct aspectual heads. Bybee (1985, 143) reports the case of Pawnee, which displays a “usitative” preverb, in addition to a habitual suffix, to mark activities occurring only “now and then” (occasionally).

38. See (aa) versus (ib), from Sterner (1975, 137, 142), who calls the first “repetitive” and the second “iterative” aspect:

NOTES TO PAGES 92-93 205

G) a. Ten-t-i-fun. REP-1|pl-r-vs. “We hit repeatedly.’ b. Y-o-rur-wer-sa.

lsg-r-vs-ITER-DIR ‘T lift up again.’ According to Capell and Coate (1984, 209), the Australian Northern Kimberley languages have “iterative” aspect suffixes “expressing the idea of repeating an “action” and “frequentative” aspect suffixes expressing “an action carried out a number of times, not just repeated once.” Here, I will use repetitive for actions repeated once (‘again’) and frequentative for actions repeated several times. See also the case of “‘semeliterative” (‘again’) and “iterative” prefixes in Navajo cited in §3.5. The term “semelfactive” is sometimes used “to refer to a situation that takes place once and once only (e.g. one single cough)” (Comrie 1976a, 42). But this seems to relate to lexical (or Aktionsart), rather than grammatical, aspect (see Smith 1991, 55ff). 39. The leftmost seems to quantify over events, the rightmost over the processes or states

represented by the verb. In literary Italian, the former but not the latter can be replaced by spesse volte (lit. ‘often times’). Also recall the observation, made in chapter 1 around (114) (Texans often drink beer versus Texans drink beer often), that the former, but not the latter, can act as an “adverb of quantification.” As pointed out by Hans Obenauer (personal communication), the two “base generation” positions of souvent ‘often’ also seem to behave differently with respect to “quantification at a distance” in French, though not all speakers perceive a clear distinction, apparently. I tentatively assume that “n times” adverbs belong to the same class as ‘often’. See the two positions of twice discussed in chapter | (§1.4). 40. Differently from spesso ‘often’, raramente ‘rarely’ seems to quantify only (or preferably) over events. Compare (43b) with *?Gianni spesso esce con la stessa persona raramente ‘Gianni often dates the same person rarely’ and *?Giulio é stato invitato raramente da Gianni, non da noi “Giulio has been invited rarely by Gianni, not by us’. I interpret the fact that the

sentence becomes possible with raramente in VP-final position (Gianni é stato invitato da tutti raramente ‘Gianni has been invited by everybody rarely’) as due to the leftward movement of [invitato da tutti] around a higher (event-quantifying) instance of raramente. Al. This is analogous to other cases of multiple occurrences of the same adverb. In (i), for example (from Vendler 1984, 306), the very same lexical item acquires (partially) different senses (“speech act,” “subject-oriented,” and “manner’”’) as a function of its base generation in different, specialized, positions: Honestly, he, honestly, told the story honestly. What remains to be seen is how a single (more abstract) meaning can yield the different senses in interaction with the semantic contribution of each scope position. In this direction, see Ernst (1987) and Rochette (1990). 42. Spesso, often, and so on can apparently occupy an additional position, at the beginning of the sentence (Spesso, Gianni sparisce per una settimana intera ‘Often, G. disappears for a whole week’). This can be an effect of the subject and the verb stopping below the higher, event-quantifying, position of spesso (see chapter 5, §5.1, for discussion of this possibility). But it may possibly also quantify over larger portions of the clause. In this connection, (107) of chapter 1 (?John twice intentionally knocked on the door) could contain twice in this higher position (hence to the left of the volitional modal adverb intentionally), with John topicalized to its left (see the text below (116) of chapter 1).

43. Bybee (1985, 150) mentions the fact that a “verbal marker that gives iterative or repetitive meaning [= ‘repeatedly’ rather than ‘again’] to the verb was found in fifteen of the 50 languages of the sample,” but does not discuss its order in the language relative to the other suffixes. 44. All this suggests the existence of two distinct quantificational “spaces”; one involv-

ing quantification over events, located just below modals, and comprising the habitual,

206 NOTES TO PAGES 93-95

repetitive(1) and frequentative(I) aspects; the other involving quantification over the predicate, comprising the repetitive(I1) and frequentative(D) aspects. A comparable distinction will be made for “quickly/rapidly” (and the so-called celerative aspect). See §§4.14 and 4.26. 45. A similar observation is made 1n Moltmann (1990) concerning the German pair... weil Hans schnell das Buch gelesen hat *... because H. has quickly read the book’ versus ... well Hans das Buch schnell gelesen hat *.. . because H. has read the book quickly’. 46, Gia can also have extended “scalar” usages (Ha letto gia cinque libri “He already read five books’), which I will not consider here. See Michaelis (1991) and van der Auwera (1993), and references cited there. In chapter 1, following much literature, J took (no?) yet as the negative counterpart of already, occurring in the same position as already. Indeed, its presuppositional

contribution is also one of temporal priority. To say that At 5, John had not yet arrived is to assert that At 5, John had not arrived and to presuppose that John had not arrived at a time before 5 either (when he was perhaps expected to arrive), Doowaayaayo is reported to have an already/yet verbal suffix (see le ‘he ate’; le-d “he already ate’; Wiering 1974, 44). A suffix glossed

as already is also found in Tuyuca (Barnes 1994, 336). A “not yet” verbal affix is found in Nkore-Kiga (Taylor 1985, 156), in the Amerindian language Amuesha (Wise 1986, 616), and in Warao (Osborn 1967, 46). On the so-called yer- and stil/-tenses of Bantu, see Dahl (1985, 176) and in particular Comrie’s (1985, 54) discussion of the not yet-tense of Luganda. 47. Even in the case of (54a~b), this is nothing more than an implicature, which can be reversed, with no contradiction (Gianni ha amato molto Maria, e l’ama ancora ‘G. has loved M. a lot, and he still loves her’); Gianni ha mangiato, ma sta mangiando di nuovo °G. has eaten, but he is eating again’). With accomplishments, which involve a natural (rather than an arbitrary) end point, reversing the implicature seems to yield a contradiction: Gianni ha mangiato il sandwich, *ma lo finira domani | e lo sta ancora mangiando ‘G. has eaten the sandwich, but he will finish it tomorrow / and he is still eating it’. With them, the perfect seems to acquire a “perfective” (or “completive’’) value, for which see the discussion in §4.23. This additional completive value is not part of the meaning of the perfect, however, as we see from the fact that 1t can be cancelled by an adverb: Gianni ha parzialmente mangiato il sandwich e lo finira domani/e lo sta ancora mangiando ‘G. has partially eaten the sandwich, and he will finish it tomorrow / and he is still eating it’. In this connection, it is noteworthy that two perfect forms exist (a “noncompletive” and a “completive” one) in languages such as Japanese (Ikegami 1985), Tamil (Herring 1988), and Hindi (Singh 1991), among others. On the similar case of Chinese, see $4.23. 48. Some Bantu languages use a special form of the verb, sometimes called “no longer tense” (in addition to a “still” and a “not yet tense”). See Comrie (1985, 54ff), and notes 46 and 50 here. 49. Both paraphrases in (55d), which contains a verb of achievement, are acceptable only under a habitual or frequentative reading (“he stopped reaching the summit regularly, every time, etc.””). Needless to say, the interaction of (non) piu with verbs belonging to the different aspectual classes would deserve a more careful scrutiny. In certain cases, the adverb gives rise to ambiguity. Gianni non canta piu can either mean that he was singing and that he has stopped, or that he was going to sing, but then decided not to. In most Veneto varieties, two different adverbs correspond to these two interpretations (pia and altro, respectively). 50. This is apparently supported by the fact that in Luganda (see (1) from Comrie 1985, 54), and in Nkore-Kiga (see (11), from Taylor 1985, 160), negating the “still tense” yields the “no longer tense”’:

(4) a. Mu-kya-tudde. you-still-sit “You are still seated.’

NOTES TO PAGES 95-96 207

b. Te-mu-kya-tudde.

not-you-still-sit ‘You are no longer seated.’

(ii) a. A-ki-rwaire. he-still-ill ‘He’s still ill.’ b. T-a-ki-rwaire.

neg-he-still-ill “He’s no longer ill.’ “A suffix glossed as ‘still’” is also found in the Sino-Tibetan language Garo (Bybee 1985, 143). A “still” (and “yet’’) auxiliary is reported to exist in Karen (Jones 1961, 17). 51. Continuative aspect is found expressed by particles, as in the Atchin example (ia) and the Banggi (Austronesian) example (1b), or (apparently, more often) by reduplication of the verb stem, (see Bybee et al. (1994), who give the Gugu- Yalanji example (ic), among others):

(i) a. Ko m’ok wiel. (Bybee et al. eds. 1994, 161) but CONT walk ‘He kept walking.’ b. Sia kahal ng-korikng [m-korikng]. (Boutin 1991, 12) is STILL ST-dry ‘Tt is still dry.’ c. yirrka-n-yiirka-y. (Bybee et al. eds. 1994, 166) ‘Keep shouting.’

Still (and continuative aspect) appears to be compatible only with nonperfect forms (see Bertinetto 1994c, 117). For a semantic analysis of the core, and extended, usages of still, see L. Michaelis (1993). 52. As ancoracan have “focusing” usages, in which it forms a constituent with the phrase following it, perhaps more revealing are cases such as (ia—b):

G) a. ?Arrabbiato con me, non lo sarai pili ancora, spero! ‘Angry with me, you will no longer still be, I hope.’ b. *Arrabbiato con me, non Io sarai ancora pit, spero! The fact that in languages with terminative and continuative verbs (rather than suffixes or particles) one can have the terminative followed by the continuative (Gianni smise di continuare a bere ‘G. stopped continuing to drink’) is not particularly telling, as the opposite order is also possible (Gianni continua a smettere di bere (senza sucesso) ‘G. continues to stop drinking (without success)’). In this case, the rigidity of syntax (assuming the order of functional projections to be rigidly ASDjemnnative > ASPcontinuativer 2S Suggested by the fixed rela-

tive order of the specifiers) is not due to the rigidity of the relative scope of the semantic operators involved. On the relation between semantics and the hierarchy of functional projections, see the discussion in §6.3. 53. In other contexts, however, it seems to be able to precede it (Quando lo vai a svegliare, Gianni é sempre ancora addormentato ‘When you go and wake him up, G. is always still sleepy’). Once again, this suggets that sempre ‘always’, like the other quantificational

adverbs seen earlier, can also be generated in a higher scope position quantifying over the event (also see n. 12 of chapter 1). It seems that in the higher position it comes after di solito ‘generally’ and before raramente ‘rarely’. Gj) a. Gianni vede di solito sempre raramente i suoi parenti. ‘G. usually always rarely sees his relatives.’ b. *Gianni vede sempre di solito raramente i suoi parenti. c. *Gianni vede di solito raramente sempre i suoi parenti.

208 NOTES TO PAGES 96-97

54. It may also be interesting to note that sempre in Italian has a usage in which it is synonymous with ancora ‘still’ (Lavori sempre all’ Olivetti? ‘Are you still (lit. ‘always’) working at the Olivetti?’). This is perhaps not unrelated to the contiguity of the specifiers hosting the two adverbs. 55. See chapter 3, n. 30. Occasionally, this aspect is called “immediate past,” a term that appears inappropriate on various counts. First, as Dahl (1985, 27) notes, it makes one think of “metrical” or “remoteness” distinctions (such as “hodiernal,” “hesternal,” etc. pasts—see n. 28 here and Comrie 1985, chap. 4), while it is found in languages “that do not otherwise mark remoteness distinctions systematically”). See also Dahl (1983, n. 2). The same observation is made in Comrie (1985, 94). Second, it is not strictly “past,” but, if anything, “anterior,” as it can refer to a time point just prior to a future time. See the Haitian example (i):

(i) Jan ap fék lévé. J. FUT RETRO get up ‘J. will have just got up.’ 56. The corresponding French adverb a peine can, in fact (somewhat redundantly), cooccur with the venir de periphrasis (// vient a peine de terminer son discours ‘He has just finished his speech’). There are differences among the various adverbs. Lately, for example, can appear preverbally only under very restrictive conditions. See Stock (1973, n. 6). Recentemente, aS opposed to appena, requires the reference time to be equal to ‘now’ (Quando arriverai, sarod appena/*recentemente partito “When you arrive, I will have just/recently left’). For discussion, see also Aqvist, Guenther, and Rohrer (1977). To retrospective aspect are possibly related such adverbial phrases as a short while ago / poco fa and a short while earlier / poco prima, which, however, due to their phrasal nature, cannot occur (in Spec) in the functional portion of the clause, but only in the VP “space” of circumstantial adverbials (or as “adverbs of setting”; $1.2). An apparent exception is the phrase da poco ‘from little’, in Italian, which has the same distribution of appena (which necessarily follows sempre):

(i) a. Gianni ha (di solito) sempre appena/da poco ricevuto qualche regalo. ‘G. has (usually) always just/from little received some present.’ b. *Gianni ha (di solito) appena/da poco sempre ricevuto qualche regalo. ‘G. has (usually) just/from little always received some present.’ 57. Once again, I gloss over the differences existing among the different adverbs. In Italian, presto is multiply ambiguous, corresponding to ‘soon’, ‘early’, and ‘quickly’ (according to the position it occupies). See also §4.25. Tra poce / dopo poco ‘In a while / after a while’ seem to belong to the same class, and can, like da poco, appear preverbally, despite their phrasal nature (Gianni sara tra poco arrivato a casa *G. will have arrived at home in a short while’; Disse che sarebbe dopo poco arrivato a casa ‘He said he would have arrived at home shortly after’). An affix (-dak-) meaning “immediately” is found in the Chadic language Deghwede. See Frick (1978, 35). 58. Possibly, the adverb improvvisamente ‘suddenly’ also belongs to this class. According to Potet (1992), the Classical Tagalog prefixes ka- and sa-, when prefixed to a measure adverbial, translate respectively as *... ago’ and ‘in...’ (e.g. kamakalawd ‘two days ago’, samakalawa ‘(with)in two days’); when they are prefixed to a verb, ka- expresses the “immediate past” (rendered with just), and sa- expresses the “immediate present” (rendered by him as ‘suddenly’ ). Jmprovvisamente, like subito and presto, also seems to follow sempre (Gianni si é sempre improvvisamente eclissato ‘G. has always suddenly disappeared’ versus *Gianni si € improvvisamente sempre eclissato ‘G. has suddenly always disappeared’). A SUDDENLYaspect suffix is reported to exist in the South American language Wapishana (Tracy 1974,

NOTES TO PAGES 97-99 209

124) in Greenlandic Eskimo (Dahl 1985, 95) and in the North American languages Quileute (Bybee 1985, 100), and Eastern Pomo. (McLendon 1996, 507); an auxiliary meaning ‘suddenly happened’ is attested in the Southern Bantu language Xhosa (Louw 1987, 12). See also the compound verb construction in Turkish meaning ‘suddenly’ (Kornfilt 1997, 36), and n. 76. 59. The same ambivalence is apparently found in the prefix -poh- of the Oceanic language Big Nambas, which Fox (1979) labels “proximity aspect’: “the action is comparatively near in time, whether past or future’ (p. 64). See example (ia—b) and (28a) of Appendix 2:

(Gj) a. N-p’sh-ma. I/realis-PROXIM-come ‘[ have just come.’ b. P’e-pah-vra-ma. V/irrealis~-PROXIM-pl-come “We shall soon come.’ 60. A subclass of such adverbials are those “implicitly related to completion (in 2 hours)” (Smith 1991, 69), and variously called (‘Frist’ [Binnick 1991, 307], “completive” [Smith 1991, 69]). They are in complementary distribution with for adverbials, being typically selected in the presence of telic predicates (“accomplishments” and “achievements”), as opposed to for adverbials, which are selected with “activities” (and certain “states’”). That for and in adverbials belong to the same semantic class is further suggested by the fact that in various languages (e.g., Chinese, Navajo [Smith 1991, 69], and Mokilese [Chung and Timberlake 1985, 237]) they have exactly the same form, acquiring their different interpretation as a function of the context in which they appear.

61. See also Svantesson (1994, 272), who notes that in Kammu (Austro-Asiatic) “the habitual [particle kz] is not used for generic statements.” 62. “Prospective aspect is indicated by an Auxiliary Verb Phrase in which g# manifests the Auxiliary element and is affixed as a Simple Verb in the normal way and -n#g is affixed to the Verb Stem which precedes the Auxiliary. This construction indicates that the commencement of the situation is imminent’ (Davies 1981, 31):

(i) Ar-n#g g-ab-in. go-purpose do-PRES-|sg ‘Lam about to go.’ 63. Li (1991, 48) analyzes the particle yuav which follows the progressive particle taabtom in Hmong (Sino-Tibetan) as future, but glosses it in the present as “getting ready to” (an aspectual more than a temporal characterization):

G) Nwg taabtom yuav yog tug thawjcoj. 3sg PROG FUT[PROSP] be CL chief ‘S/he is in the process of getting ready to be a chief.’ Uhlenbeck’s (1938, 179) discussion of the verbal prefix dk(s)- in the North American language Blackfoot is particularly telling: “It is decidedly no sign of the future tense, but only an adverbial element with the meaning about to, going to, nearly, or the like”:

Gi) Aksikoko. ‘It was nearly night (“it was going to be night’). See also Hollingsworth’s (1991) analysis of the particle da following the progressive in MofuGudur (Afroasiatic), discussed in n. 3 of Appendix 2. Although it must be distinguished from ordinary future, both semantically and positionally, it may still share with it some property at a more abstract level. Steele (1975, 46) renders the Thai particle, cd?, with “future” when to the left of the progressive particle kamlang (see (ilia)), and with a prospective periphrasis when to the right of kamlang (see (iib)). This was confirmed by Pornsiri Singhapreecha (personal communication), who notes, however, that for her (iila) is marginal without a time frame:

210 NOTES TO PAGES 99-101

(iii) a. Khaw ca? kamlang pay. he FUT PROG go ‘He will be going.’ b. Khaw kamlang ca? pay miawaannii.

he PROG PROSP go yesterday ‘He was going to go yesterday.’ 64. In fact, as already noted in n. 39 of chapter 3, Heine (1992, 339) labeled this aspect, manifested in various african languages, the ALMOST-aspect. In Hunzib (Caucasian) the “suffix -a.Ae denotes that an action is almost happening or is almost accomplished” (van den Berg 1995, 111): A.u-l akar r-ol.l-a.A e-r ‘(S)he almost finished the sugar’. See also the ALMost suffixes of the Australian language Kayardild (Evans 1995, 261), and of the North American language Mandan (Mixco 1997, 30). In Vietnamese, the same morpheme (gdn) is rendered in Thompson (1965, 215) through a prospective periphrasis with activity verbs (ia), and as ‘almost’ with stative predicates (ib):

GQ) a. Tol gan di Sai-gon. ‘I’m on the point of going to Saigon.’ b. Téi gan muén khoo. ‘Tam almost sick.’ In the Papuan language Tauya, “the auxiliary /-wa?ase-/ implies that the state or action specified by the verb nearly took place: pomu-wa rase-e- ra ‘Inearly fell’” (MacDonald 1990, 205). In the Austronesian language Banggi, the aspect auxiliary adak, rendered by Boutin (1991, 14) as ‘almost’, “is used to refer to a situation . . . that the speaker believes was possible but that did not actually occur” (Adak ku me-dabu’ “Almost I fell’). In Northern Pomo (an Amerindian language) “there is a suffix, -s’u, which may occur immediately before the [perfective] suffix -ye. It indicates that the state of affairs denoted by the verb almost took place, but did not.”: Mito xa:kanam-su-y “You almost drowned’. To judge from Edo, which possesses both a particle (rd), glossed by Omoruy (1991, 12) as ‘to be about to’, and one (té), called “unfulfilled aspect” and glossed as ‘almost’ (p. 13), Northern Pomo -s’u ‘almost ..., but did not’ is possibly an aspect to be kept distinct from prospective aspect. 65. Not relevant here is the additional possibility for guasi to directly modify completamente (in its Spec). The two readings differ in their intonational contour. 66. Just as completamente and tutto are quite low in the adverbs’ hierarchy, so is completive aspect in the heads’ hierarchy. In chapter 3, I noted that the evidence (available to me) was not sufficient to determine the relative order of Completive aspect and Retrospective, Progressive, and Prospective aspects. The order of the corresponding adverb classes (specifiers) provides, then, the evidence missing from the heads’ side. In this analysis, tutto, like completamente and the other AdvPs, is in an A-bar position. Given its QP status, it binds from there a variable (the A-position where it was “base generated’). See Cinque (1992a) and references cited there. In fact, there is evidence (discussed in §5.3) that the position occupied by tutto is not the same position (plausibly, an A-position) occupied by floating quantifiers. 67. If no object is present (as in intransitives), or if the object is indefinite, it is the plural subject which falls under the scope of the completive operator. See (ia—b), from Tepehua, a language spoken in Eastern Mexico (Watters 1988, 229ff). In the same language (see (1i)) the null object of a transitive verb with a completive suffix is rendered as ‘everything’ as in Wayampi (Tupi-Guarani); see (111), from Jensen (1994, 349):

(i) a. Nis’at’an-n ta-‘u-‘o:-lni pu:laqti. Art child-Pl 3SubPl-eat-COMPL-PEREF art tamale ‘The children ate all of the tamale(s).’

b. Nis’at’an-n ta-‘u-‘o:-1 pu:laqti Art child-P] 3SubPl-eat-COMPL-PEREF tamale ‘All the children ate some tamale(s).’

NOTES TO PAGES 101-103 211

(ii) Lage’in-oho-y. see-COMPL-IMPERF *X sees everything.’ (iii) N-o-me’é-ma-i ajama’é. Neg-3A-give-COMPL-Neg Cntr:Exp ‘But (counter to expectation) he did not give everything.’ 68. But not all Polish speakers agree with this judgment. The two “completive auxiliaries” found in Ainu are possibly another case in point. See example (ia~b), from Refsing (1994, 316):

GQ) a. Cepcie wa tsam. fish we eat PERF COMPL “We ate up (all) the fish b. Toon korsi tu pon cep e wa okere. that child two small fish eat PERF COMPL ‘That child finished (the process of) eating the two small fish.’

69. Intransitives that have internal stages and a natural end point are compatible with completamente: Hai digerito completamente? “Have you digested completely?’ 70. The case of bene + passive past participle discussed in the text should be kept distinct from the (lexically restricted) truncated variant ben + past participle, actually possible even with active past participles (see La valigia é stata ben sistemata ‘The suitcase has been well placed’; L’hai ben sistemata? “Have you well placed it?’), which appears to be a case of incorporation (left adjunction) of the adverb to the verb (see Rivero 1992). Evidence to this effect, brought to my attention by Richard Kayne (personal communication), is provided by absolute participial clauses like (1), where the past participle is independently known to move to C® (here with the adverb). See Kayne (1989a, 97); Cinque (1995, 231 n. 24); Belletti (1990, chap. 2): (i) Una volta ben sistemato anche te, tuo padre si rilassera (versus ?? bene sistemato.) ‘Once well placed you too, your father will relax.’ 71. Maria Cristina Figueiredo informs me that similar facts hold in Brazilian Portuguese. The same has been found to hold in Greek (Alexiadou 1997, 154f). The discussion here has tacitly assumed that raising of the verb to different heads is motivated by the requirement of checking the marked value of one (or more) functional head(s). This question, which opens the possibility of relating different interpretations of the verb to the different syntactic positions it comes to occupy, is very delicate and requires careful investigation. Since French past participles presumably display the same interpretations as Italian past participles even if they do not move (as much) in overt syntax, the weak/strong feature parameter of Chomsky (1995) must also hold, to explain V-movement orthogonally to the marked/default value assumed here. The order bene/etc. + passive past participle is due to the past participle having stopped in a head to the right of the adverb, rather than to a (special) movement (or “base generation”) of the adverb to the left of the past participle. This is indicated by the fact that the relative order of tutto and bene remains the same, and by the fact that, accordingly, the passive past participle can follow tutto as well. See (i), and Cinque (1992a):

(i) a. Quando sara stato tutto bene sistemato, .. . ‘When will have been everything well arranged, .. .’ b. *Quando sara stato bene tutto sistemato, ... “When will have been well everything arranged, .. .’

72. A potential problem is provided by the apparent possibility of coordinating bene and presto (Lo tradussero bene e presto / presto e bene ‘They translated it well and quickly/

212 NOTES TO PAGES 103-106

quickly and well’). Adverbs belonging to (positionally) distinct classes should not be able to be coordinated. The sentence final position of the second AdvP opens the possibility that this 1s not a coordination of AdvPs, but of larger constituents, with a reduced second conjunct: Lo tradussero bene e [lo tradussero] presto. This is confirmed by the impossibility of coordinating the two AdvPs when some other constituent follows (which rules out the coordination of two larger constituents): *Lo tradussero bene e presto / presto e bene in turco ‘They translated it well and quickly / quickly and well into Turkish.’ 73. I thank Richard Kayne for pointing out to me the contrast between (85) and (86). As he also noted, there are certain examples (belonging to a different style) in which fast can precede the finite verb (He fast ran out of money). Erich Groat has noted that -/y-less adverbs can occupy only the lower position (*He quick ran away).

74, “Celerative -law-, -ilaw- ... indicates speed of action, or earliness, and is sometimes used in combination with the adverbial form jaawd’um ‘quickly’, e.g. ‘o-mabbilawiinde ‘he shut it quickly’ (Arnott 1970, 356). “CELERATIVE. The affix is -(2)law-, and it adds to the verbal concept an idea of rapidness. . . . Interestingly, Jaw is also the phonological form of the adverb meaning ‘fast, quickly’. The derived forms look like these: . . . Mi nyaam-lawan ‘I (will) eat quickly’’(Fagerli 1994, 36f). Verbal suffixes glossed as ‘quickly’ are also found in Mandan (Siouan, North America; Mixco 1997, 30), in Dyirbal (Australian; Dixon 1972, 248), in Evenki (Caucasian; Nedyalkov 1994, 1331997, 252), and in the Sino-Tibetan language Boro (see n. 79 here). A verbal suffix glossed with ‘early’ is found in West Greenlandic (Fortescue 1984, 279ff), and in the Ama-

zon language Axininca Campa (Payne 1981, 42). Aleut has different suffixes for ‘fast’, ‘slowly’, and ‘quickly’ (Bergsland 1994, 366). Gamble (1978, 54) calls “retardative” a particular verbal aspect of Wikchamni (Penutian, North America), whose function “is to modify the inherent activity of the bases to a slow, extended action.” 75. Iabstract away here from the further (derivational) suffixes (or “radical extensions”; Arnott 1970, sections 57-59), that can be interspersed between the REPETITIVE, COMPLETIVE,

and CELERATIVE aspect suffixes. They appear related to grammatical functions (or argument positions): reflexive, associative, instrumental, benefactive, reciprocal, and so on. 76. Judging from the Ika case just mentioned, this aspect 1s higher than Progressive aspect. Possibly, it is to be distinguished from Inchoative aspect, which marks “the coming about of a state” (Smith 1991, 35). Less clear is what class of adverbs these aspects are related to.

Possible candidates could be adverbs of the “suddenly” class, unless they are related to Proximative aspect (as tentatively suggested in n. 58 of this chapter). See Schlyter (1977, 76): “Certains adverbes (la serie a: brusquement, etc.) indiquent qu’ un événement commence, et on peut les appeller INCHOATIFS.”

77. The Conative aspect suffix is located in Tauya (Papuan) between the Habitual and the Perfect aspects. See (33a—b) of Appendix 2. This aspect, the Terminative, Continuative, Inceptive, Success aspects seen here, and the Predispositional aspect (“tend to”) found in American Sign Language (Klima and Bellugi 1979), seem to correspond to some of the “aspectual verbs” of Romance which allow for Clitic Climbing (see Rizzi 1978, Kayne 1989c). While I do not have sufficient evidence to determine their relative order, they seem to fall in between Habitual aspect and Voice (see Appendix 2). 78. Oksapmin (Indo-Pacific) is another language with a suffix (-ptt) that “expresses a sequential relationship between the first and second unit” (Lawrence 1972, 34). A particle indicating that ‘‘an action is first of a series of related actions” is found in the African lansuage Tikar (Jackson 1980, 35). A “priorative” particle meaning “to do something first, that is, prior to something else” is also found in Yapese (Jensen 1977, 213). See also the suffix -gqaar of West Greenlandic (Fortescue 1984). To judge from the Bantu language Kako (U. Ernst 1991, 38), which has an analogous aspectual particle “a valeur prioritaire,” this func-

NOTES TO PAGES 106-109 213

tional head can be generated in different positions along the hierarchy (acquiring different scopes), in a way possibly similar to focusing adverbs. 79. Bhat (1994, 76) notes that “in Boro ..., there are very few independent adverbs ... the function of adverbs is usually performed by a number of verbal suffixes. Bhat (1968) records a total of about 260 suffixes of this nature that can be attached to verbs.” Some illustrative examples of “adverbial” suffixes attached to the verb za ‘eat’, given by Bhat (1994, 77) are: zaglé ‘to eat quickly’; zabay ‘to eat again’; zagaw ‘to eat possibly’; zakma ‘to eat stealthily’; zaka ‘to finish eating’; zakri ‘to begin to eat’; and zaké ‘to pretend to eat’. A similar situation is apparently found in Garo (Sino-Tibetan) and Chemehuevi (Uto-Aztecan) (Bhat 1994, 76). Schachter (1985, 23), in addition to the case of the North American language Yana (for which see also Sapir 1921, 126), mentions Eskimo, which “has a large set of suffixes with adverbial meanings” (like -nirluk ‘badly’, -luinnag ‘thoroughly’, -karik ‘unfortunately’, -qquugq ‘probably’, etc.). The same is true of the closely related West Greenlandic (Fortescue 1984, 301) and Aleut (Bergsland 1994, §4). 80. In chapter 6, I consider whether a single, more abstract (semantic), principle could be at the basis of these two distinct syntactic conditions, arguing that this would still not do. Note, incidentally, that in other areas of grammar absence of overt morphological realization of a certain grammatical notion is not automatically taken as evidence that the notion has no syntactic realization (or role). For example, though nonpronominal DPs show no morphological Case distinctions in English or Italian, we generally do not conclude from that that no Case is present in sentences with nonpronominal DPs. Similarly, as Holmberg and Platzack (1995, 179, n. 35) note, one would not conclude from the fact that North Swedish verbs lack a present tense form distinct from the infinitive that “this dialect does not even have T/TP in sentences interpreted with present tense.” It seems to me that we have no reason to adopt a different attitude in the case of other functional heads and projections.

Chapter 5 I wish to thank Paola Beninca’ and @ystein Nilsen for discussing with me different parts of this chapter, and Anna Cardinaletti, Richard Kayne, and Martin Prinzhorn for their reactions to an earlier draft. For the data discussed here, unless otherwise indicated, I am indebted to Carmel Coonan (British English), @ystein Nilsen (Norwegian), Verner Egerland (Swedish), and Halldor Sigurdsson (Icelandic). 1. In any case, it seems necessary to replace a purely Case-related approach to DP positioning with a more articulated one. If more positions are available to DPs outside the VP, presumably they will not reduce to Case requirements only. 2. A possible way out of this conclusion would be to posit a movement of the verb in PF which could possibly do without the presence of a head position. But, if the different position of the verb relative to the various AdvPs is not semantically innocent (see chapter 6), then the presence of a verb between the DP, or FQ, and AdvP, cannot be a purely PF phenomenon. 3. Under the assumption that adverbs are (left) adjoined to (functional) X°’s (see Travis 1988), one would naturally expect that between two contiguous adverbs (.e., two contiguous X°’s) there could be a DP (an XP); which is what we find. One, however, would not immediately expect a verb (1.e., another X°) to intervene between the DP and the second adverb.

But such cases are entirely natural (in Italian): Forse gli uomini non sono necessariamente degli esseri sociali ‘Perhaps men aren’t necessarily social beings’. 4. Baker (1971, n. 3) and Jackendoff (1972, 76) regard (2) as ill-formed, noting however that some people accept it (in fact, Richard Kayne informs me that for him it is accept-

214 NOTES TO PAGES 109-110

able). Perhaps, for those who accept it, either probably can be used as a focusing adverb (see $1.6), or have (av) is actually a reduced form of of, a complementizer, in Kayne’s (1997) analysis. 5. See §2.3. In some sense, multiple generation/adjunction of adverbs is made conceptually redundant by the possibility of deriving their different positioning from the independently necessary DP- and V-movements. The theory developed here can be seen as offering a principled reason to elimimate the redundancy. 6. Similar facts hold in Spanish, modulo the absence of an adverb corresponding to mica (see Lois 1989, 23ff).

7. The position to the left of mica is also a “scrambling” position for PPs, in Italian (Gianni non haa Maria mica piu detto niente, poi “G. has not to M. not any longer said anything, then’). See Kayne (1975, 65, n. 81; 1994, 75) on similar facts in French. Note that the ungrammaticality of (3b—o) cannot be due to the fact that the finite verb is too low, lower than (some) “lower” adverbs. Example (ia—l), in which the same adverbs precede the finite verb (but, crucially, follow the subject) are grammatical, at least in more careful styles of Italian (see Appendix 1). In no style, however, can the finite verb follow bene ‘well’ or presto ‘early’; see the ungrammaticality of (¢m—n) (all the adverbs here are intended without focus intonation. For differences between this option and focalization, available to a subset of the adverbs, see §2.2):

G) a. Di solito Maria mica prende i treno. ‘Usually M. not takes the train.’ b. Per le una, Maria gia sara di ritorno. ‘By one o’clock, M. already will be back’ c. Maria pili non mi pensa. ‘M. no longer thinks of me.’ d. Maria ancora gli parla. ‘M. still speaks to him.’ e. Maria sempre ripete le stesse cose. ‘M. always repeats the same things.’ f. ?Maria appena si era coricata, quando squillo 11 telefono. *M. just had gone to bed, when the phone rang.’ g. Maria subito mi avvertiva, quando non veniva. ‘M. immediately would call me, when she wasn’t coming.’ h. Maria brevemente ci sta parlando della sua avventura. ‘M. briefly is telling us about her adventure.’ i. Maria quasi cadde dall’emozione. ‘M. almost fell for the emotion.’ 1. ?Maria completamente distrusse tutto quello che aveva fatto fino ad allora. ‘M. completely destroyed all that she had done till then.’ m. *Maria bene fece tutti 1 compiti. *M. well did her homework.’ n. *Maria presto si alzava ogni mattina. *M. early would get up every morning’

For evidence that in (i) the adverbs occupy their usual position, and that it is the verb which occupies a lower position, see the end of chapter 2. Italian, then, is a little closer to English than is usually assumed. In both languages, the finite verb raises obligatorily to a head to the left of ‘well’ (and ‘early’), with Italian, but not English, further allowing it to raise to higher positions in overt syntax (see *He recovered completely early versus He completely recovered early and He [recovered early] completely

NOTES TO PAGES 110-113 215

t). This is schematized in (41) (V and * indicate possible and impossible positions of the verb, respectively):

(ii) a. *frankly *luckily *allegedly *probably *then *perhaps *necessarily “willingly *obligatorily *wisely *usually *again *often *quickly *already *no longer *still *always *just *soon *ailmost *completely Vv well *early b. Vfrancamente Vfortunatamente Vevidentemente V probabilmente Vallora Vforse Vnecessariamente Vvolentieri Vobbligatoriamente | saggiamente Vdi solito Vdi nuovo Vspesso Vrapidamente Vgia V(non) piti Vancora Vsempre Vappena Vsubito Vquasi Vcompletamente Vbene *presto

Norwegian essentially patterns with English. The finite verb has to precede godt ‘well’ and (a fortiori) tidlig ‘early’ (see (iiia) and (iiid—e); it can either precede or follow helt ‘com-

pletely’ (see (iiib—c); and it must follow ali higher adverbs (Oystein Nilsen, personal communication):

(ii) a. *...at han ikke lenger alltid helt godt sov. *,.. that he not any longer always completely well slept.’ b. ... at han ikke lenger alltid helt sov (sa) godt. c. ... at han ikke lenger alltid sov helt godt.

d. *... at han tidlig vaakner opp. ‘that he early wakes up.’ e. ... at han vaakner opp tidlig. 8. As Anna Cardinaletti noted (personal communication), the possibility of having the “weak” (subject) pronoun egli ‘he’ in (4), which resists (Clitic) Left Dislocation, suggests that the nonpronominal subjects of (4) are not necessarily (clitic) left dislocated. 9, The contrast between Icelandic (i) and Norwegian (i1) (from Jonas 1996, 181) would follow if bad were inserted in the Spec of a higher DP-related projection and det necessarily in the lowest, which could relate to the strong nature of AGRs (Vangsnes 1995), or to the generalized nature of V-to-I(-to-C) of Icelandic (Vikner 1995), much in the spirit of Holmberg’s (1993, §9) proposal (Italian, however, patterns with Norwegian, despite its strong AGRs):

(i) Pad hafa nokkrar kékur verid bakadar fyrir veisluna. ‘There have some cakes been baked for the-party.’

(ii) a. *Det har noen kaker blitt bakt for selskapet. “There have some cakes been baked for the-party.’ b. Det har blitt bakt noen kaker for selskapet. “There have been baked some cakes for the-party.’ c. Noen kaker har blitt bakt for selskapet. ‘Some cakes have been baked for the-party.’

The possibility for Norwegian (and Swedish) to have (definite and indefinite) subjects in between various classes of adverbs in the “middle field” in nonpresentational sentences (see (7) in the text) could relate to the lack of overt subject agreement on the finite verb, which would dispense with the need of pre-Spell-out Spec/head agreement, hence with the insertion of a null expletive. For general discussion, and an interesting analysis, of these data, see Vangsnes (1995, forthcoming). 10. The partial acceptability of (6h) is likely related to the possibility for soon to occupy a very high position as well (possibly, that of “adverbs of setting’’), to the left of probably (see Soon the train will probably stop). {1. As @ystein Nilsen points out, “subject-oriented” adverbs like klokelig ‘wisely’ do not seem to allow the subject to follow them (*... at klokelig Per gikk hjem ‘.. . that wisely

216 NOTES TO PAGES 113-114

P. went home’). This may have to do with their “subject orientation.” Perhaps they need to be predicated of the subject already in overt syntax. In subordinate clauses the (finite) verb has to precede godt (and all adverbs lower than that), and may precede helt, but no higher adverb, while the facts concerning the subject remain the same. English finite verbs (and auxiliaries, including modals) also have to precede well, and all adverbs lower than that (apart from cases like As you well know... ), and may remain lower than completely. On the position of object DPs, see §5.2. 12. Languages (even closely related ones, such as Norwegian and Danish) apparently differ as to the positions they make available to the subject. In Danish, for example, “adverbials never occur to the left of the subject” (Vikner 1995, 128). See also Holmberg and Platzack (1995, 181), Afarli (1995). Full DPs in Danish thus behave like pronominal subjects in Norwegian (and some other Germanic languages), which have to precede all adverbs. In this respect, Swedish goes with Norwegian (Platzack 1986, 43ff; Holmberg 1986, 98,1993, 31ff and n. 3), and Icelandic with Danish. (“In Icelandic, it is never possible to have an adverbial preceding the subject” (Vikner 1995, 106). See also Sigurdsson 1989,43; Holmberg and Platzack 1995, 18). This, however, does not hold for indefinite (or definite but contrasted) subjects, which can occur, even in Icelandic, after various classes of adverbs (also in what are called Transitive Expletive Constructions). See (i) (Holmberg and Platzack 1995, 181, cite an additional exception):

(i) a. [ ger klarudéu sennilega margar mys ostinn. (Bobalijk and Jonas 1996, 196) “Yesterday ate probably many mice the cheese.’ b. I fyrra bordudu alltaf margar mys ostinn (Halldér Sigurdsson, personal communication) ‘Last year ate always many mice the cheese.’ c. [ bessum bidum gatu sennilega sjémennirnir keypt allt bad sem burftu. ‘In these shops could probably the SAILORS buy all they need.’ (Vangsnes, forthcoming, ex. 43b) Bobaljik and Jonas take TEC subjects to occupy Spec,TP. The fact that they follow alltaf ‘always’ and all adverbs higher than that (Halld6r Sigurdsson, personal communication) shows that they occur quite low in the functional structure of the clause, however. As noted, after Halid6ér Sigurdsson (personal communication), indefinite subjects cannot easily follow reet ‘just’ and adverbs lower than that (‘briefly’ or ‘almost’ [see (tia—b)], nor ‘completely’, ‘well’, or ‘early’):

(ii) a. *bad hafa stuttlega margir menn lyst mali sinu. ‘There have briefly many people described their case.’ b. *had hodeféu naestum margir menn komi0 timanlega. ‘There had almost many people arrived timely.’ The “extraposed” sentence-final subjects found in languages with TEC, and in those without, I take to be in Spec, VP. 13. Recall that mica is just above gid ‘already’ in Italian, and that English apparently lacks an element corresponding to mica; this renders the two languages potentially identical with respect to the position to which the subject raises obligatorily. 14. The dependence of such interpretive differences on the different position of the subject was originally noted for German by Angelika Kratzer (Diesing 1992, 33; see also Kratzer 1989, 12). Whether the two distinct positions of the subject are necessarily linked to the individual-level/stage-level nature of the predicate, as claimed by Diesing, is less clear (for discussion, see Moltmann 1990 and Kiss 1996, among others). For similar observations concerning Dutch indefinite DPs, see Reuland (1988) and references cited there.

NOTES TO PAGES 114-116 217

15. The ja doch adverbs in (9) and (10) seem to be very high in the adverb hierarchy, as they precede evaluative adverbs (such as leider ‘unfortunately’), and all adverbs lower than that. Gerhard Brugger, who provided the relevant judgements, notes that (ja) doch can also follow leider (and a number of adverbs lower in the hierarchy), with a concomitant change in interpretation, changing from a speech-act adverb qualifying the assertion to a presuppositional adverb which reverses a negative expectation (comparable to Italian si ‘yes’, which is located much lower; see chapter 1, n. 20). Besides Diesing’s, other recent works suggest the existence of more than one subject position, differently specialized. For example, Kiss (1996) argues for two VP-external subject positions in English comparable to Diesing’s distinction in German; Cardinaletti (1996) argues for two specialized subject positions in Italian, a Case-related one, and a higher one for subjects of predication (weak pronominals being able to fill only the lower of the two). Similarly, Egerland (1996) proposes a PersonP (higher than Agr,P), to the Spec of which pronominals, but no full DP, can raise (see also Haegeman 1993, 149). Oystein Nilsen notes that the weak quantifier noen ‘somebody’ in Norwegian can be interpreted existentially after uheldigvis ‘unfortunately’, but must be interpreted as specific when to its left. This observation, and the earlier one concerning ja doch may help one to narrow down the location of the Spec relevant to “specificity.” 16. Another question is whether the different DP-related projections are always structurally present or whether they are “created” only when needed (to host a DP in their Spec). Perhaps there is no uniform answer, and scope-sensitive projections are present only when needed, while interpretively specialized ones (for “existential” or “distributive” readings) are always present. For further discussion, see chapter 6 on the marked/default value of functional heads. 17. As he points out, it is not always easy to construct natural examples with any two contiguous adverbs (e.g., with tydligen ‘obviously’ and sannolikt ‘probably’, or with ofta ‘often’ and redan ‘already’). Also, concerning (15a), the preferred option is with drligt talat topicalized: Arligt talat gjorde jag lyckligtvis det inte ‘Frankly did I luckily it not’. 18. In the literature, it is often claimed that Norwegian (Mainland Scandinavian), as opposed to Icelandic, allows “object shift” of pronominals only, not of full DPs (see, among others, Holmberg 1986; Vikner 1994, §7; Holmberg and Platzack 1995, 141ff; Diesing 1996b, 75). To judge from (17)—(18), that conclusion is inaccurate. A possible reason, pointed out to me by @ystein Nilsen, for why “object shift” of full DPs was not noticed in Norwegian is the fact that the testing cases have typically involved the negative adverb ikke (inte, in Swedish), which indeed does not allow full DPs, or VPs, to raise across (although it allows pronominals):

(i) a. Per liker den, ikke t,. ‘P, likes it not.’ b. *Per liker Jon, ikke t,. c. *Per har [likt Jon]; ikke t,. As Nilsen notes, (ib) becomes much better if ikke is stressed contrastively:

Gi) Per liker Jon IKKE. The well-formedness of (17e) can be reconciled with the il!l-formedness of (ib) if ikke in (17e) is in the Spec of lenger. This is indirectly confirmed by the fact that autonomous ikke precedes allerede (just as mica precedes gid in Italian: Per har vel ikke allerede kommet! ‘P. has really not already come’, while the ikke of ikke lenger follows allerede (Per elsker allerede ikke lenger sin kone ‘P. already no longer loves his wife’). Crucially, as pointed out to me by Nilsen, nothing can intervene between the latter ikke and lenger. See: *Sin kone elsker allerede ikke Per lenger i det hele tatt ‘His wife P. already no longer loves at all’. Although the phenomenon underlying (ib) and (ii) remains to be understood, the adverbs

in (17)-(18) show that “object shift” of full DPs is possible. That it is “object shift” rather

218 NOTES TO PAGES 116-117

than VP shift is shown by the fact that at least some of the adverbs that allow the full DP object to precede the adverb do not allow VP shift (@ystein Nilsen, personal communication): *Per har [sovnet] heldigvis/HELDIGVIS ‘P. has fallen.asleep fortunately’. For general discussion, see Nilsen (1997). These facts confirm the conclusion, based on Faroese (Vikner 1994, 502) and Old English (Roberts 1995, 281) that the trigger of “object shift” is not the checking of morphological Case (but see Holmberg and Platzack 1995, §6.5.6). They also show that “object shift” should not be related to generalized V-to-I movement (Vikner 1994, 502ff). 19, Full object DPs, as well as pronominals, also occur interspersed among “higher” adverbs in German (see Brugger and Poletto 1993 and Haftka 1995 and references cited there).

The fact that topicalized and focalized full DPs cannot precede a lexical complementizer (Haftka 1995), and that no more than one of them can precede the verb in V/2 contexts, may be taken as an indication that the complementizer is in Force° (in Rizzi’s 1997 terminology), or higher, and that the verb raises to Finite® (or Foc®). For similar conclusions, see Grohmann

(1997) and Poletto (1997). 20. In particular, following Shionsky (1991) and Giusti (1991, 1993, chap. 4), I take DP to be the complement of Q, raising in this construction, to Spec,QP and out. 21. For this observation, see Sportiche (1988, 435). 22. I owe these judgments to Marie-Therése Vinet (personal communication). 23. Sportiche (1988, 435) takes the interpretation of (21) to be with the FQpo preceding the FQjo, attributing the observation to Richard Kayne. This may be a typographical error as Kayne (1975, 156) interpretes it with the FQ; preceding the FQpo. As Kayne notes, this at least holds for FQ; not introduced by 4. 24. Sportiche (1992, §7.3) claims that the unmarked order is with the FQ,, preceding the FQgs, which in turn precedes the FQpo. This claim appears incompatible with the two separate claims made in Sportiche (1988, 435) to the effect that the FQ, precedes the FQpo, and the FQno precedes the FQio (see the previous footnote). For, on that basis, by transitivity one should expect the FQ, to precede the FQ,y We have already seen that the FQ{,> must be taken to precede the FQpgand that the FQ, must also be taken to precede the FQpo, so the question reduces to whether the FQ, precedes the FQ,o, or vice versa. A possible way to reconcile the apparently contradictory judgments reported in the literature is to take the unmarked order in the FQ “field” to be FQ, > FQio > FQpo, as done in the text here, and to assume an extra (“scrambling”) position to be available to the FQ, (with a) to the left of the FQ “field.” See the beginning of n. 7, and the following discussion for independent evidence to this effect concerning Italian loro. 25. The distribution of the FQ, a tutti in Italian appears to be that of an ordinary a+full DP, rather than that of a FQ. The same appears to be true in French as well. See Kayne (1975, §2.14).

26. Cardinaletti (1991, 150), who notes the same contrast between the orders loro > tutti and tutti > loro, still finds the order loro > tutti rather marginal, unless tutti is focalized (n. 21), which she interprets as implying that tutti has remained in the object base position. To my ears, (22a), though awkward, is unexceptionable, even if tutti is not focalized (focalization of the “inverted” subject excludes that tutti be focalized, given that no more than one focus is possible in Italian; see Calabrese 1984). 27. Examples G) and (ii) differ from (25) only in the location of the past participle with regard to the FQ, and loro. It appears that the best location for the past participle is that between the FQ, and loro:

(1) a. Han sempre tutte dato loro tutto. they (FEM) have always all(FEM) given to-them everything b. *Han sempre loro dato tutte tutto. they (FEM) have always to-them given all(FEM) everything

NOTES TO PAGES 117-121 219

(ii) a. ?Han sempre dato tutte loro tutto. they (FEM) have always given all(FEM) to-them everything b. *Han sempre dato loro tutte tutto. they (FEM) have always given to-them all (FEM) everything

28. Napoli (1974, 483) gives such cases as I ragazzi le hanno tutti viste “The boys (MASC) them (FEM) have all (MASC) seen (FEM)’, where an FQ, precedes a past participle agreeing with a clitic object, as ungrammatical. But I find them quite acceptable (see also the analogous cases in French discussed earlier in the text). I also differ from Cardinaletti (1991, n. 22) in finding (i) (comparable to Cardinaletti’s Gib)) “awkward” but essentially acceptable with the appropriate intonation (with a preference for the location of the past participle in between the two quantifiers; see (11)): Gi) Le paste, 1 ragazzi le hanno apprezzate tutti tutte poco. ‘The cakes, the children them have appreciated all all little.’

(ii) Le paste, 1ragazzi le hanno tutti apprezzate tutte poco. ‘The cakes, the children them have all appreciated all little.’ 29. See the S > IO > DO order of “scrambling” positions in West Flemish (Haegeman 1993).

30. We continue to ignore the post participial location of either tous or tout when they are focused:

Gi) a. Les enfants ont tout lu (presque) TOUS. ‘The children have everything read (almost) ALL.’ b. Les enfants ont tous lu (presque) TOUT. “The children have all read (almost) everything.’

31. These judgments are Marie-Therése Vinet’s (personal communication). 32. The almost acceptable status of Han capito tutto tutti bene (‘They have understood everything ali well’), alongside the preferred Han capito tutti tutto bene, could be due to the fact that tutti forms a constituent with bene (See TUTTI BENE, li ho fatti ‘Ail well (focus), I made them’; Sono riusciti? Si, tutti bene “Have they come out all right? Yes, all well’). This option is apparently not available to loro. (See *?Spiego tutto loro bene Gianni ‘G. explained everything to-them well’). 33. Apparently, in the presence of (ne) plus, the position of FQ’s after toujours becomes unavailable:

(i) a. *Ils n’ont plus toujours tous tout fait. b. *Je ne les ai plus toujours tous invité. c. *Je ne leur ai plus toujours tous tout dit. 34. The FQ may, however, come to precede (ne) pas in L-tous cases. See (ia—b), from Kayne (1975, 23, n. 25):

(i) a. ?Elie aurait tous pu ne pas les prendre. ‘She would have all been able to not take them.’ b. ?Elle aurait pu tous ne pas les prendre. ‘She would have been able all to not take them.’ 35. The head status of the negative element in NegP1, which precedes the finite verb, and the phrasal status of the negative elements in NegP2, NegP3, and NegP4, which generally follow the finite verb, is motivated in Zanuttini’s study by the fact that the former, but not the latter, appear to block verb movement. See the following text for further discussion.

220 NOTES TO PAGES 121-122

36. This is, for example, the case with the Piedmontese variety spoken in Cairo Montenotte, which can simultaneously lexicalize NegP1, NegP2, and NegP4, with the effect of a single negation (Dy ’menika une pa ‘vnynu ‘He didn’t come on Sunday’). See Parry (1997, 181) and Zanuttini (1997, chap. 3). (Quebec) French provides evidence for an additional negation in the COMP “space”: J’ai caché les ciseaux pour pas que ma fille se blesse ‘[hid the scissors so that my daughter wouldn’t hurt herself (Daoust-Blais and Kemp 1979), a possibility apparently available only in varieties with postverbal negation, as noted by Richard Kayne and Raffaella Zanuttini in unpublished work. Other Romance varieties (e.g., those of Cembra, Lisignano, and Brazilian Portuguese) appear to have a very low negation, following all complements (see Zanuttini, 1997, §3.4 and references cited there), That, however, may be the effect of raising to its left what follows one of the higher NegPs, much as was proposed in §1.4 for “lower” VP-final adverbs. 37. That French guére is possibly in the same slot as Milanese and Pavese no is indicated by the fact that it follows toujours and precedes complétement (see chapter 1, n. 19). 38. If NegP2 (the presuppositional negation) can fail to count as the “real” negation, one can make sense of its co-occurrence with the “real” negations, in NegP1 in Italian and in NegP3 in certain varieties of Piedmontese (see the pa nen of (46) of Zanuttini, 1997, chap. 3). NegP1 can also fail to be the “real” negation, in which case it co-occurs with NegP3 (and NegP2), as in the Piedmontese of Cairo Montenotte (see n. 36), or NegP2, as in French. Interestingly from this perspective, no Romance variety seems to lexicalize NegP3 and NegP4 simultaneously. Perhaps, neither of them can fail to be the “real” negation. 39. This implies a different treatment of the different positioning of the two negations with respect to the subject and complement clitics. Perhaps, the “real” negation cannot be a two-segment category ([y [neg cl neg J V ]); hence it must be the last element to adjoin ([y [a neg cl] V |). Lassume that direct cliticization to the verb is in fact compatible with antisymmetry requirements (for discussion, see Cinque 1996, §6). 40. Paola Beninca’, who provided the relevant judgment, notes that subject clitic/verb inversion is not possible if the conditional is counterfactual (*No fusse-lo ‘riva’ in tenpo, gavarissimo vudo problemi ‘Had he not arrived in time, we’d had problems’). I have no explanation for that. 41. In Paduan, a finite verb preceded by negation cannot apparently raise to C® in literal interrogative clauses (1), though it can in certain exclamative and nonliteral interrogative clauses (ii) (see Beninca’ and Vanelli 1982; Zanuttini, 1997, chap. 2):

(i) a. *No ve-to via? (see No te ve via?) ‘Don’t you go away?’ b. *Cossa no ga-lo fato? “What hasn’t he done?’ (ii) a. Cossa no ghe dise-lo? “What doesn’t he tell him?’ b. Vien-lo o no vien-lo? ‘Does he come or not?’ c. No vien-lo miga? ‘Doesn't he come?’ Zanuttini (1997) and Portner and Zanuttini (1996) propose that, when it contributes the ordinary meaning of negation to the clause, no is an independent head (of NegP1), which can be attracted to interrogative C° (instead of the verb), at least in yes/no questions. It is instead cliticized to the verb, making a unit with it, when it contributes an implicature (as in (41)), whence the position of negation + V to the left of subject clitics. Although such an analysis is in principle compatible with that developed here (the facts in (52) only show that the “real” negation can cliticize to the verb, not that it must), 1t is possible that the ungrammaticality of (i) may depend on some other factor. Perhaps, the “real” negation cliticized to the V prevents the interrogative feature on the V from entering the required Spec/head agreement (Rizzi 1991) with the interrogative C°.

NOTES TO PAGES 122-124 221

42. The type of contrast between (53g) and (54c) is noted in Ladusaw (1978, 1988). 43. The reading with forse ‘perhaps’ under the scope of negation requires some contrast. For example, it is favored in the following context: ... /’ha fatta di sicuro! ‘. . . he/she made it for certain!’. Here, forse contrasts again with probabilmente, which cannot fall under the scope of negation even in such a context. 44. The two interpretations correlate with a slightly different intonational contour. When it is outside the scope of negation, the adverb appears to be stressed more prominently. When under the scope of negation, the adverbs differ as to whether they admit being alone under the scope of negation (In questo paese, i negoziantinon rilasciano obbligatoriamente ricevute; solo se ne hanno voglia ‘In this country, shopkeepers do not obligatorily issue receipts; only if they fancy it’) , or only as part of a larger constituent (Gianni non era di solito disposto a compromessi; *lo era sempre ‘G. wasn’t usually disposed toward compromises; *he always was). The latter cases are only paraphrasable as “Tt is not true that ... (G. was usually disposed ...).” As Paola Beninca’ notes (personal communication), this difference correlates with the possibility, or impossibility, for the adverb to be directly modified by negation (Non necessariamente questa clausola avra conseguenze ‘Not necessarily will this condition have consequences’ versus *Non di solito Gianni era disposto a compromessi ‘Not usually was G. disposed toward compromises’). Deliberatamente, obbligatoriamente, spesso, rapidamente, and sempre pattern with necessariamente; astutamente, di nuovo, gid, and quasi pattern with di solito. When mica, or another negative element, is present below the adverb, only the reading

with the adverb outside the scope of negation seems to be possible (Non é di solito mica disposto a compromessi ‘He is usually not disposed toward compromises’ ). If mica is higher than the adverb (e.g., Gianni non aveva mica accolto sempre le loro offerte °G. had not always accepted their offers’; Gianni non riceveva mica gia notizie da casa ‘G. did not already receive news from home’), the adverb is under the scope of negation, whether alone (sempre), or as part of a larger constituent (gid). 45. See Schlyter (1977, 100, 215) for an analogous observation concerning the corresponding French adverbs. Holmberg et al. (1993, §2.5) note a comparable contrast between epistemic adverbs and “lower” adverbs in Finnish. They take epistemic predicates to (obligatorily) raise at LF across negation (noting a problem for this idea in English). This is dubious, however. Overt movement of adverbs across negation is generally blocked (*How often haven’t

you seen him? *COMPLETAMENTE, non lo ha revinato! ‘Completely (focus) he hasn’t ruined it’) (and when it appears possible: How hasn’t he behaved!? it acquires a special rethorical status). Where it is systematically allowed, as in the special topicalization construction discussed in Cinque (1990, §2.4), the adverb obligatorily “reconstructs” under the scope of negation (Completamente, non lo ha rovinato ‘He hasn’t ruined it completely’). Similarly, in German, “l’adverbe est aussi sous la negation s’il est en téte d’une phrase principale, en portant l’accent le plus fort de la phrase. Ex. Héflich hat er nicht geredet |’ He hasn’t spoken kindly’]” (Schlyter 1977, 38, n. 1). Against LF movement of adverbs, see also T. B. Ernst (1991, 754 ff; 1992).

If the “base generation” position (rather than a derived LF position) is crucial for this type of scope, then such cases as (1) (= example 61a) of Holmberg et al. (1993) in Finnish, in which “potential mood obligatorily has scope over negation,” have a derivation more similar to that of Italian non, with e (neg) starting in some lower Spec and ultimately left-adjoining to the agreement head 7:

(i) Pekka ei liene unohtanut sinua. P. neg-agr be-POT forgotten you ‘P. has probably not forgotten you.’ 46. A case of the same general sort is provided by an example discussed in §4.7 (John must probably give his money back), which showed the scope epistemic modality > deontic

222 NOTES TO PAGES 124-125

modality in spite of the surface order in which the deontic modal must precedes the epistemic adverb probably. Once again, this can be made sense of if must is generated below the adverb, and is raised above it via head raising, and if it is the “base” position which determines the relative scope of the two elements. For similar conclusions, see T. B. Ernst (1991, 754 ff; 1992, §4). As Richard Kayne reminds me, the “surface” position of negation is instead crucial for licensing polarity items (also see Laka 1990, §3.2.2.1):

(i) a. Why didn’t anyone help us? b. *Why did anyone not help us? 47. This requires, as noted, that it be possible to cliticize (left-adjoin) something (here, the negation) to a verb (see Cinque 1996, $6). English would minimally differ from Italian in having n’t (not) generated in the head, rather than in the Spec of NegP, and picked up by the (auxiliary) verb, which left-adjoins to it on its way up (see Kayne 1989b). Alternatively, n’t is already attached 1n the lexicon to the verb, which then raises to the relevant head to check the negative feature. 48. A crucial element in the derivation of the facts in (54) and (57) 1s the fact that adverbs, in the normal case, do not move (from lower positions to their “surface” position). See chapter 1. If they did, their scope interaction with negation would be more complex. In this connection, it is interesting to note that (quantified) DPs, which—we know—move up within the sentence, have a different behavior. As noted in the literature, quantified subjects can be interpreted under the scope of negation even if they precede negation at Spell out. See Tutti i giocatori non hanno segnato ‘Every player didn’t score’ and Jous ne sont pas venus ‘All didn’t come’. The fact that they can also take scope over negation is in line with the scope properties of A-movement, which (in contrast to head movement) allows either the head, or the foot, of the chain to count for scope (See A hippogryph is likely to be apprehended and the relevant discussion in May 1985, 97~105 and McCloskey 1996b, 14). 49. The reading in which both deliberatamente and di solito are outside the scope of negation is more clearly brought out if mai ‘never’ is inserted (between di solito and polemico). 50. If the highest NegP position available is indeed between Mod, .crem< and T(Past), the fact that “if a language is verb-initial [VSO or VOS] the negative will precede the verb” (Dryer 1988, 97) suggests that the V does not raise to COMP. Dryer cites only one language (Cariri), out of 53, as contradicting his generalization, but this exception seems apparent only, as the negation in second position is a suffix on the verb (see (1)), so that it becomes possible to take the verb to left-adjoin to a negative head in first position in this case too:

(i) netso-kié di-dé i-fia. be.seen-NEG his-mother him-by ‘His mother has not been seen by him.’ Colloquial Welsh at first sight provides another counterexample to Dryer’s generalization, with a sole postsubject negation in certain cases. But see Payne (1985, 225) for evidence that a preverbal negation always exists, reinforced by a second postsubject negation. 51. In some languages, the negative suffix can appear in different positions among other suffixes (acquiring different scopes). This is the case of the negative suffix -ri- of Tuyuca (Tukanoan, South America), which “negates only the information which occurs to its left” (Barnes 1994, 331). See (ia—b) (= examples (26) and (27) of Barnes 1994):

(i) a. Bué-ruku-ri-wi. study-constantly-NEG-EVID. ‘T did not study constantly.’ G.e., “I studied but not constantly.’) b. Bué-ri-ruku-wi. study-NEG-constantly-E VID. ‘I! constantly did not study.’ (i.e., ‘I was constant 1n not studying.’)

NOTES TO PAGES 125-131 223

52. Incidentally, this crosslinguistic variability can itself be construed as an argument to set negation apart from the other adverb-related functional elements, which do not appear to show variability, as noted. To judge from the typological works of Dahl (1979), Payne (1985), Dryer (1988), and Kahrel and van den Berg (1994), sentence negation can appear as either (a) a suffix (see the Turkish case in (63a) and the case of Tuyuca in the preceding footnote), (b) a prefix (see the case of Berber in §6.4), (c) a clitic on the verb (Italian), (d) a negative auxiliary (see the Finnish case of n. 45), (e) a particle (e.g., the Standard Arabic invariant morpheme maa; Fassi Fehri 1993, 162ff; Ouhalla 1993), (f) an AdvP in Spec (e.g., Norwegian and Piedmontese). Examples a, d, and e involve a morpheme in X°, either bound (a), or free (d and e); f involves a phrase in Spec; b and c, the head of a phrase in a subjacent Spec incorporating to a V, possibly in the morphology (in the case of prefixes), or in the syntax (in the case of clitics). 53. This sentence is (irrelevantly) acceptable as a topicalized version of Non ha parlato brevemente ‘(He) has not spoken briefly’ (with the adverb in the scope of negation). 54. That the ungrammaticality of the sentences in parantheses, under (64a) and (64b), is due to the presence of negation, not to the verb having stopped to the right of the adverbs, is shown by the grammaticality of Gf) and Gh) of n. 7 here. 55. The conditions that allow more than one (generally “lower’) NegP to count as a single negation are another matter, which remains to be understood (see n. 38).

Chapter 6 For comments on some of the issues raised in this chapter, I am indebted to Paola Beninca’, Anna Cardinaletti, Alessandra Giorgi, Erich Groat, Richard Kayne, Giulio Lepschy, David Pesetsky, and Fabio Pianesi. |. The original notion of markedness is due to Nikolaj Trubetzkoy (Waugh 1982, 300), and was later elaborated by him and Jakobson. For recent discussion and references, see Battistella (1990, 1996). 2. As Paola Beninca’ suggested to me, the apparent “ambiguity” of the default value in Jakobson’s formulation (either unspecified or the opposite of the marked value) might be elegantly resolved if conceived of as an “elsewhere” case. Assuming the marked value to be [-fortunate], the default value, ‘-[-fortunate]’, indeed covers both ‘fortunate’ and ‘unspecified for the opposition fortunate/unfortunate’. This suggestion suits the other cases as well, and is particularly illuminating for the volitional modal case mentioned in the next footnote. 3. This choice of marked and default values appears to be supported by the volitional/ non-volitional morphology found in a number of languages (see chapter 4, n. 34). In such languages, the morpheme marking volitionality is used both for events that involve volition and for those that cannot involve volition (such as ‘rain falling’). This suggests then that {-volition]) is the marked value, and that its opposite (-[-volition]), which covers both cases (see, again, the previous footnote) is the default value. This observation appears to generalize to the other modal heads. 4. In many Creoles (and in some African languages), the unmarked tense is apparently “present” in generic and stative sentences, but “past” with eventive (non-stative) sentences. See Déchaine (1991; 1993, 250f) and the references cited in DeGraff 1996, 125). 5. The fact that “small clauses” allow for the full range of AdvPs, including the high-

est (Con Gianni francamente inadatto a questo compito .. . “With G. frankly not suitable to the task ...’), suggests that structurally they may be full clauses (with a lexical projection distinct from VP). Independent considerations for the full clausal nature of “small clauses” are found in Cinque (1995b), Sportiche (1995a), and Starke (1995), and references cited there.

224 NOTES TO PAGES 131-136

6. See Ouhalla (1988, 74; 1990, 207ff) and Poletto (1993, 295), who, however, take the temporal and aspectual auxiliaries, not the tense and aspect bound morphemes, to be generated under the corresponding functional heads. 7. As Richard Kayne notes, He HAS always admired her is however possible as a contradiction. 8. As Richard Kayne pointed out to me (personal communication), auxiliaries appear in some Cases (in certain languages) to take complements larger than VPs (can be followed by what looks like a complementizer), Here, as in the previous chapters, I have abstracted from the evidence pointing to the existence of a P/CP projection dominating the projection(s) occupied by participles in Romance (see Kayne 1993b). In any event, I take such P/CPs not to “close off” the extended projection of the lexical V, but to be part of it. 9. The equivalent with guickly is possible (... and he quickly has/will), but recali that quickly (as opposed to fast) can also occur in a higher position (where it quantifies over the event). 10. For example, no “dangerous”-aspect seems to be attested, in spite of its potential significance for us (the adverb exists as a member of the much wider class of circumstantial (manner) adverbials, whose general meaning is “in an X way’). Likewise, no language appears to have a mood expressing the fact that there is ‘bad weather’ (or ‘good weather’), though we can easily imagine a world where that would be possible. Examples of this sort could be multiplied. 11. This is bound, of course, to be controversial. Sabine Iatridou once asked if we really want to generalize to all languages a “politeness” functional projection (presumably, a DP-related one; see §3.1), for which we have morphological evidence (based on verbal suffixes) only in few languages. Although I do not remember what I answered at the time, now | think I would answer “yes.” Politeness, and deference, distinctions (both with respect to the arguments of the predication, and to the addressees) are apparently found universally (even if only few languages mark them via affixes on the verb). See Comrie (1976b), Brown and Levinson (1987). 12. Their assumption is that UG makes available a universal stock of functional fea-~ tures, much like the universal inventory of phonetic features, from which individual languages pick different subsets. However, this may be seen as projecting the accidental gaps of the lexicon and the morphology into gaps of the underlying system—a move, as noted, which is not followed in other areas (for example, with Case or phi-features). For this reason, I would

take the analogy with phonology to involve, not particular choices of phonetic features, but the CV skeleton, syllable structure, and so on, which are necessary components of all languages. 13. Recall that with accomplishments, the default value of the completive aspect head is [+completive], so that even without the adverb (John has eaten the sandwich) the interpretation is that he ate it completely. 14. I abstract away here from the additional “optional” possibilities seen in chapter 2. 15. Of course, one must establish precisely at which particular functional heads each specific verbal form must, and can, raise in each variety. Although relevant observations are to be found in chapters 2 and 5, in Pollock (1989), and in Zanuttini (1997, chap. 3), a systematic investigation remains to be done. 16. As Richard Kayne pointed out to me, other cases are less acceptable, however: */t is probable that what I'm told is that John will give up. 17. Of course, the co-occurrence of functional elements may well be affected by their semantics. That is, certain combinations may be “filtered out” on purely semantic grounds. For example, it is plausible to think that the incompatibility between Retrospective Aspect and non anterior Tenses (*He will recently laugh) arises from purely semantic incompatibilities. This, as well as other such cases, deserve a separate study.

NOTES TO PAGES 137-139 225

18. This does not exclude the possibility that there may be prevalent tendencies in the positioning of negation or agreement. For example, from Bybee’s (1985, 35) study, it appears that (person) Agreement suffixes are more frequently outside Tense and Mood suffixes than vice versa (in other words, that agreement is marked more frequently on a DP-related projection higher than the Mood and the Tense projections than in a lower one). 19. Itis not entirely clear, however, that the difference between the two languages should be so interpreted. Note that a further difference must be assumed: that heads Jeft-adjoin in Turkish, but right-adjoin in Berber, to the next higher head (the latter possibility being, in fact, unavailable in the more restrictive antisymmetric framework). Much depends, in any event, on the proper analysis of prefixation, which is not yet fully understood. If prefixation in Berber were of the Navajo type (see §3.5), with successive raising of a null auxiliary to the various functional heads, followed by cliticization of the result to the main verb ([Aux]-negtense-agr#Verb), the two languages would have exactly the same underlying structure (note that the linear order of the affixes indeed is the same in the two languages: neg-~tense-agr). This analysis, however, appears dubious, given the possibility for clitics to intervene between Neg and Tense. Compare (17a) in the text with G) here:

(i) Ur-tn ad-y-agh. (= example (27a) of Ouhalla 1988, 40)

NEG-them will-agr-buy ‘He will not buy them.’ Perhaps the negative element left-adjoins to the clitic, and the other heads cliticize (phonologically) to the main verb. I leave the question open. 20. In fact, she provides evidence for the existence of other functional projections, from which I abstract here. These host either degree elements (viak, pal ‘right’) or er proforms. The presence of the former elements, in particular, interacts in compiex ways with PP internal movements. 21. For ease of exposition I put the lexical elements directly in Place® and Path’. See Koopman’s 1993 article for more careful discussion. 22. Limited cases of head-to-head incorporation are found in English, however, where such (Place+Path) compound prepositions as into and onto are plausibly derived by leftadjoining the lower (Place) head in/on to the higher (Path) fo. 23. In this, I simply follow a suggestion of Riemsdiyjk’s (1996), who however takes the suffixes to be the heads of head-final projections. 24. In addition to the further functional structure discussed in Koopman (1993), it remains to be seen how best to represent more complex cases like From out of the cold (through the tunnel) into the smoky room (he went) (see Jackendoff 1973, 351). Perhaps, PathP should be further split into distinct projections (of Origin, Path and Goal). 25. Recent work suggests that the same may be true of the CP “space,” which appears to be composed of different projections functionally specialized—Brody (1990), Vikner (1991), Hoekstra (1992, 1993), Miiller and Sternefeld (1993), Watanabe (1993), Alber (1994), Platzack (1994), and Rizzi (1997), among others. On the basis of Romance and Germanic, Rizzi (1997) proposes the hierarchy of projections in (i): (2) LForceP Force [Topicr Topic LeocusP Focus Leiniter Finite lip te

This order appears to have wider cross-linguistic validity, with topic phrases systematically preceding focalized phrases. ForceP may be embedded in another CP layer, of pure subordination (see Bhatt and Yoon 1991, Rizzi 1997), and some evidence suggests the existence of an InterrogativeP, distinct from FocusP (between this and FiniteP). See the Hungarian evidence discussed in Puskas (1996), and such cases as (ii) in Italian, where a focalized object co-occurs with, and precedes, an interrogative phrase (something possible only in embedded contexts):

226 NOTES TO PAGES 139-145

(ii) a. Mi chiedo GIANNI chi possa aver osato invitare! ‘I wonder G. (focalized object) who (subj) can have dared invite.’ b. *?Mi chiedo chi GIANNI possa aver osato invitare!

In fact, the COMP “space” may turn out to be richer still. See Beninca’ (1996) for evidence of a wh-exclamative CP, apparently higher than TopicP. See also Puskas (1996) and Poletto (1997).

Appendix 1 J. Actually, if we consider lower “lower” adverbs (like tét ‘early’), we see that even in French infinitives have to raise a little (but I will not pursue this question here):

(i) a. *Il faut tot se réveiller. ‘One must early wake up.’ b. fl faut se réveiller tét. ‘One must wake up early.’ 2. As noted in chapter 1, after Pollock (1989, 413), there are speakers who allow the infinitive even to raise past plus (this is also true of one author (Montherlant) cited in Engver 1972, 22ff), but they appear to be a minority. It is interesting from our perspective that such oscillations among speakers occur at the “cutting points” of verb movement on the adverb hierarchy; not at any lower point (with the effect, say, that some speakers, but not others, allow the infinitive to the left of bien, while all speakers allow it to the left of toujours). 3. See also Rizzi (1982, 103). The fact that in such construction the infinitive can follow forse (La persona a cui forse rivolgersi in tali circostanze’ ‘The person to whom perhaps

to refer under the circumstances’) and adverbs higher than forse, but must precede necessariamente (La persona a cui rivolgersi necessariamente/? ?necessariamenie rivolgersi in tali circostanze “The person to whom to refer necessarily/ necessarily to refer under the circumstances’ and adverbs lower than necessariamente) appears to provide evidence for the obligatory raising of the infinitive to Mood, ,eans. 4. But see n. 80 of chapter 1. Lonzi (1990, 142ff) mentions some cases where the infinitive of a subject clause is preceded by an adverb, as in meramente annuire (é da bambini) ‘to merely nod (is childish)’, but this possibility seems restricted to “focusing” adverbs, which can modify any constituent (see §1.6). To judge from Ojea Lépez (1994, 414), infinitives (in subject and object clauses) in Spanish also raise obligatorily past adverbs which can remain to the left of finite verbs (see section 4) and (3):

(i) a. Protestar siempre no te va a ayudar. (see *Siempre protestar . . .) “To always protest will not help you.’ b. Todos queremos marcharnos ahora (see * .. . ahora marcharnos)

Lois (1989, 43) gives comparable examples as marginal rather than ungrammatical. 5. Zanuttini (1997, chap. 4, n. 54) also notes that temporal adverbs like adesso cannot precede such infinitives. 6. Note that these sentences all involve verbs not requiring their infinitival complement to be introduced by an overt (prepositional) complementizer. As Belletti (1990, 137, n. 63) pointed out, the clitic nature of infinitival complementizers could interfere with the presence of an adverb. Indeed, in general, no material can intervene between di or a and the following infinitive (Rizzi 1982, 97ff), except in higher stylistic levels (see Cinque 1983, n. 13). In fact, given the latter possibility, it is tempting to reinterpret the ban against material intervening between the infinitival complementizer and the infinitive not as due to the clitic nature of the complementizer, but to the obligatory movement of the verb to the head immediately below

NOTES TO PAGES 145-147 227

C? (in nonstylistically marked sentences). The complement, in (9), is preposed (focalized) to prevent the preinfinitival adverbs from being interpreted as modifiers of the matrix clause. 7. When the infinitive moves to C°, in Aux-to-COMP constructions, all “higher” AdvPs expectedly follow the infinitive (and the nominative subject; see Belletti 1990, 49; 1994b, 36ff): L’aver io per fortuna/probabilmente/ora/forse/stupidamente risposto in maniera brusca ‘I fortunately/probably/now/perhaps/stupidly having answered in a brusque manner’ (versus *T,’ aver per fortuna/probabilmente/ora/forse/stupidamente io risposto in maniera brusca. See also Should he unfortunately/probably be angry versus “Should unfortunately/probably he be angry; Kayne 1984, 215). “Higher” adverbs analogously follow the finite verb moved to C° in V/2 contexts in West Germanic and in Scandinavian languages (unless they themselves move to Spec CP); see Koster 1978; Taraldsen 1983; Platzack 1986; Vikner 1995; and Afarli 1995. For the position of the subject with regard to AdvPs in Scandinavian, see §5.1. 8. Subject infinitival clauses appear to allow for dislocated phrases only very marginally. Compare (ib) with (1a) (a case of focalization):

(i) a. A GIANNI scrivere (non a Mario) sarebbe opportuno! “To G. (focus) (not to M.) to write would be in order.’ b. *?Quell’informazione, averla gia avuta ci da un vantaggio That piece of information, to have already had it gives us an advantage 9. Bellunese:

G) a. *L’ha del tut pers la testa. ‘He completely lost his mind.’ b. L’ha pers del tut la testa. ‘He lost completely his mind.’ c. L’é sempre ‘gnest. ‘He has always come.’ d. ??L’é ‘gnest sempre. ‘He has come always.’ e. No 1’é mia pi andat. ‘He has not any longer gone.’ f. *No 1V’é mia andat pi. “He has not gone any longer.’ Catalan:

i) a. *No he mai vist na Maria. ‘I haven’t never seen M.’ b. No he vist mai na Maria. ‘T haven’t seen never M.’ c. *No he pus(maiorc)/mes(maiorc/valenc) dormit. ‘IT haven’t any longer slept.’ d. No he dormit pus/mes. ‘T haven’t slept any longer.’ e. No he pas vist na Maria. ‘T have not seen M.’ f. No he vist pas na Maria. ‘T have seen not M.’ 10. Part of the data concerning Pavese, Milanese, and Piedmontese are from Zanuttini (1997, chap. 3, table 143). For the remaining facts of these varieties I am indebted to Andrea Moro, Massimo Vai, and Alda Rossebastiano, respectively. For Friulian I am indebted to Paoli (1997). For Paduan, I thank Paola Beninca’; for Trevisan, Antonietta Bisetto; for Bellunese, Nicola Munaro; for Venetian, Cecilia Poletto and Delfina Mayer. 11. Very few adverbs (mai ‘again’, si ‘even’, tot ‘still’) can in fact be found between the auxiliary and the active past participle in Romanian; but they have been argued to be clitic, like the auxiliary (see Dobrovie-Sorin 1990, 1994, 26). Following an observation by Celia Jakubowicz, Lois notes (1989, 34) that in Spanish too (certain) adverbs can intervene between the auxiliary and the active past participle, if heavily stressed (Juan ha SIEMPRE hecho lo que ha querido ‘J. has always done what he liked’). No adjacency is required with infinitival auxiliaries. See (1), from Lois 1989, 58). See also Sufier 1987, 684):

G) a. (El) no haberla nunca visto. ‘(The) not having ever seen her.’ b. (El) haberlos siempre invitado. ‘(The) having always invited them.’

228 NOTES TO PAGES 147-156

Some also allow adverbs to intervene with polysyllabic finite auxiliaries (habia, habria, hubiera, etc.). See, for example, Platero me habia ya saludado con un rebuzno ‘P. had already greeted me with a bray’ (Sufier 1987, 685), though this is not possible, apparently, with mas (*Juan no habta mas vuelto ‘J. had no longer returned’). 12. For pragmatic reasons, it is difficult to construct examples with speech-act adverbs (francamente ‘frankly’, etc.). These examples show that no adjacency between the past participle in C and the subject is required. The ungrammaticality of the adverbs in the (c) examples, if no other constituent follows appears related to their impossibility in postfinite V position when no other constituent follows (*Gianni arriva fortunatamente ‘G. arrives luckily’). They become possible if “nght-dislocated.” As shown by (28c), an exception is represented by speech-time adverbs (see §1.1). Their well-~formedness in final position may be related to their occurrence also as VP-final adverbs (see §1.2). 13. The fact that the (b) examples are perhaps slightly more natural may suggest that movement to C is preferred even in the control construction. 14. Possibly due to its presuppositional content, mica is incompatible with subordinate clauses not carrying the primary assertion of the sentence (in Erteshik-Shir’s 1973 sense); whence the ungrammaticality of: *(Non) mica arrivato sempre puntuale, Gianni si scuso “Not arrived always punctual, G. excused himself’. 15. Mai, when it precedes the past participle, is incompatible with non, differently from piu (see (ia). This is possibly related to its being an inherently negative element (differently from piu—see chapter 1, n.10).

G) a. (*Non) mai arrivato puntuale in vita sua, quella volta Gianni sorprese tutti. b. Non arrivato mai puntuale in vita sua, quella volta Gianni sorprese tutti. ‘Not having ever arrived punctual in his life, G. that time surprised everybody.’ 16. Present participles in French appear instead to obligatorily precede all adverbs. See Kayne (1975, 20ff). 17. But see n. 80 of chapter 1. In more careful styles of Italian, finite verbs can however remain much lower, following completamente ‘completely’ (and all adverbs higher than completamente), though not bene ‘well’ nor presto ‘early’ (see chapter 5, n. 7, and relative text).

Appendix 2 1. -la is a (direct) evidential past, versus say (a plain past, morphologically identical to the anterior tense). -Ze, instead, is an (indirect) evidential past. 2. Haspelmath (1993, 145) thus defines “continuative aspect”: “The Continuative ... adds the semantic element ‘still’ (negative ‘anymore’). Sometimes the adverb hele ‘still’ is redundantly present.” 3. This reconstruction of the order of functional heads in Mofu-Gudur is quite tentative as it reinterprets some of the labels of Hollingsworth’s (1991) while at the same time

postulating movement of non-head portions of the functional structure. First, I interpret the particle da, occurring between the progressive aspect particle fd and the verb not as T(Future) but as Prospective aspect. This is supported by Hollingsworth’s own characterization of the construction: “fd and da can appear together to indicate a close future progressive situation, one which is about to occur’’(p. 250), and by such examples as (7a) in the text, which encode a “prospective” situation in the past, not the future (also see the case of Gungbe in §3.3). Second, J interpret the particles cdy (sém, sat) as T(Anterior) rather than “perfect” as they give rise to pluperfect (‘anterior of the past’) with the past tense particle ta (see (7b), and precede a particle that Hollingsworth calls “completive,” and which

NOTES TO PAGES 156-160 229

appears to mark Perfect aspect in such cases as (7c) (it is common for non-past perfective to mark the future; Comrie 1976a, 67). Finally, I take the postverbal and postcomplement position of T(Anterior) and Aspyertect to be due to the raising of a non-head constituent past them; a derivation rendered plausible by the final position of the Question particle in (7d), also stranded by leftward movement of its complement (in Kayne’s 1994 spirit). Raising of the complement of T(Anterior)/Aspperfece Past it is also common in many African languages, and in Creoles (see n. 4 of this appendix and chapter 3, n. 26; see also the case of the Austroasiatic language Kammu in (19b) here, and of the Malayo-Polynesian language Kwaio in (22b)). 4. See (i) and (ii), from Omamor (1982); see also Dahl (1985, 178). I take the V in (1) to have moved within a larger constituent to a Spec intermediate between Asp pertece ANd ASPpabitual-

If this constituent is ASPprogressiveP» it may be possible to explain the “unexpected” order in (i1), in which ASPprogressive Precedes ASPperfect:

G) Owaaékaré rén. (Omamor 1982, 110) He FUT HAB go PERF ‘He will have started going (habitually).’ (ii) M6 waa winéron (gba) ré ron. (Omamor 1982, 119) I FUT PROG go PERF ‘T will already have been going.’ 5. As it is rendered with ‘étre en train de’ (U. Ernst 1991, 32), I glossed “Imperfectif” as Progressive. What the author calls “Duratif” is here called Continuative as the author repders it with encore/toujours (p. 34) rather than ‘for a while’. 6. Thave rendered as IMPERF(ect) what Chia calls “durative” as it alternates with, and has the opposite value of, PERFECT, which marks “ceasure of an event” (Perfect aspect is incompatible with Inceptive aspect—p. 78). 7. Although Babu refers to the suffix -ITTUND in (18b) as “perfect,” I consider it a T(Anterior) since it yields, with the past, an anterior of the past. The suffix -um, which is traditionally considered Future, according to Babu may be a modal. The negation -illa can also appear before aayir (T(Past)). The suffix -unnu is generally a non-past (present) suffix. Babu does not comment on its apparent co-occurrence with Past in (18a) and (18b). See the similar case of Aleut in (16). 8. The (nonnegative) perfect particle hd6oc is found either before or after the verb and its complement, or both before and after. 9. Steele (1975) glosses cé? as “INCOMPLETE,” but by her own description it should be glossed as FUTURE. As Pornsiri Singhapreecha informs me (personal communication), the ca? which precedes the PROGRESSIVE particle kamlang is interpreted as Future; the one following it as a PROSPECTIVE aspect particle. The two interpretations are not interchangeable. 10. As Keesing notes (p. 119), the particle bi’i in combination with the future marking particle “indicates that the action of the verb will take place at some (unspecified) time a short while in the future (‘after a while’ or “‘by-and-by’).” He also renders it as ‘soon’. This corresponds to what is sometimes called “proximative aspect” (see Fox 1979 on such aspect in Big Nambas and (28) here). He also notes that “by itself it indicates recent completion of the action of the verb, and can usually be translated as ‘just’: apparently, what is referred to as “retrospective aspect” (see n. 30 and relative text in chapter 3). This may suggest that “retrospective” and “proximative” are but two values of one and the same, more abstract, aspect.

But see the discussion in chapter 4 (§4.19). Finally, I take the sentence final perfect (or T(Anterior) ?) particle no’o of (22b) to arise via movement of its complement around it (as observed in many other languages; see the end of n. 3). Lt. Progressive aspect is expressed in Ponapean by reduplication of the stem.

230 NOTES TO PAGES 160-165

12. I have rendered sei as IRR(ealis mood) even though Mosel and Hovdhaugen gloss it in these two examples as “optative.” The reason is that, in their words, it indicates “that the event has not been realized yet” (p. 357), and because they gloss it elsewhere (pp. 359, 582, etc.) as “Subyunctive.” On the complex particle ‘olo’ou, see the end of n. 27 of chapter 3. 13. Vonen glosses na in (27b) as PAST, but says that it is “a marker of relative past tense” (1994, 384), and glosses kua as INCH(oative), saying however that “it must be seen in the perfective aspect” (p. 386); hence my gloss. On the systematic relation between perfective and inchoative (called by him “ingressive’’), see Comrie (1976a 19ff). 14, The Amerind unity is however quite controversial. 15. Apuis called “continuative,” butis almost always rendered with the progressive form. 16. Abbott calls “completed action” what I glossed as PERFECT. 17. Cole (1985, 197) explicitly says that the order of the causative (-chi-), prospective (-gri-) (called by him “ingressive’), progressive (-ju-) and durative (-riya-) suffixes is: (V-)

chi- > -gri- > -ju- > -riya- ... (see his (779)), which confirms the order of heads found in Guyanese Creole and Gungbe: Aspgurauve > ASPprogressive > ASPprospective

18. Although Kouwenberg (1994, 64) calls it “perfective,” I have glossed -te with T(Anterior) as she notes that with the (absolute) past tense particle ’wa’ it yields the anterior of the past:

(i) En moi jermatoko wa la-te hiso. one good woman = child PAST arrive-te here = so ‘A nice girl had come here.’ As with Sranan irrealis future particle sa, I take Berbice Dutch sa to be generated in Mood, eats;

and to raise to T(Future). See n. 33 of chapter 3. She also notes that “[r]eduplication of the verb root marks repetitive or iterative aspect” (p. 63). 19. Ofuani (1982, 229) mentions the co-occurrence of the preverbal particles go (future), don (anterior), and de (progressive), in that order. Furthermore, to judge from their relative position with regard to negation, the particle bin (past) precedes go (future). See (ia—b):

(3) a. Ai bin no de du mat wok dat taim. (Ofuani 1982, 236) I PAST NEG PROG do my work at that time ‘I wasn’t doing my work at that time.’ b. Ai no go foget. (Ofuani 1981, 315)

I NEG FUT forget ‘T will not forget.’ 20. Troike (1996, 658) gives the following order of suffix classes (noting, however, that they do not all occur together): V-pretensive-again-desiderative-negative-dubitative(perhaps)tense-interrogative. “Pretensive” appears to correspond to the “simulative” aspect suffix of Fula/Fulfulde ‘to pretend’, occurring below Voice, and similar affixes in other languages (cf. §4.28). I have rendered ‘dubitative( perhaps)’ with Mood,,.a,, (cf. $4.9). 21. Zuni is a linguistic isolate spoken in North Western New Mexico. See Nichols (1993, n. 1).

REFERENCES

Aarons, Debra, Benjamin Bahan, Judy Kegl, and Carol Neidle. 1992. “Clausal Structure and a Tier for Grammatical Marking in American Sign Language.” Nordic Journal of Linguistics 15:103-142. ————. 1995. “Lexical Tense Markers in American Sign Language.” In Karen Emmorey and Judy S. Reilly, eds., Language, Gesture and Space. Lawrence Erlbaum, Hillsdale, N_J., pp. 225-253. Abbott, Miriam. 1991. “Macushi.” In Desmond C. Derbyshire and Geoffrey K. Pullum, eds., Handbook of Amazonian Languages, vol. 3. Mouton de Gruyter, Berlin, pp. 23-160. Aboh, Enoch. 1993. Théorie syntaxique et syntaxe du Gun. Mémoire de Maitrise, University of Geneva.

———. 1996. “A propos de la syntaxe du Gungbe.” Rivista di Grammatica Generativa 21:3-56. Abraham, Werner, and Josef Bayer, hrsg. 1993. Dialektsyntax. Linguistische Berichte, Sonderheft 5. Westdeutscher Verlag, Opladen. Acquaviva, Paolo. 1995. “Operator Composition and the Derivation of Negative Concord.” Unpublished ms., University College, Dublin. Adamson, Lilian, and Norval Smith. 1995. “Sranan.” In Jacques Arends, Pieter Muysken, and Norval Smith, eds., Pidgins and Creoles: An Introduction. John Benjamins, Amsterdam, pp. 219-232. Adéwolé, Laurence O. 1989. “Sequence and Co-occurrence of Yoruba Auxiliary Verbs.” Canadian Journal of Linguistics 34:1-17. ——-——. 1991, “Aspect and Phase Systems in Yoruba.” Calgary Working Papers in Linguis-

tics 14:1-19. Afarli, Tor. 1995. “A Note on the Syntax of Adverbial Phrases.” Norsk Lingvistisk Tidsskrift 13:23-—-40.

231

232 REFERENCES

Ahn, Hee-Don, and Hang-Jin Yoon. 1989. “Functional Categories in Korean.” Harvard Studies in Korean Linguistics, Vol. 3. Department of Linguistics, Harvard University, Cambridge,

Mass., pp. 79-88. Aksu-Kog, Ayhan, and Dan Slobin. 1985. “The Acquisition of Turkish.” In Dan Slobin, ed., The Crosslinguistic Study of Language Acquisition, Vol.1. Lawrence Erlbaum, Hillsdale,

N.J., pp. 839-878. ————. 1986. “A Psychological Account of the Development and Use of Evidentials in Turkish.” In Wallace Chafe and Johanna Nichols, eds., Evidentiality: The Linguistic Coding of Epistemology. Ablex, Norwood, N.J., pp. 159-167. Alber, Birgit. 1994. “Indizi per l’esistenza di uno split-CP nelle lingue germaniche.” In Gianluigi Borgato, ed., Teoria del linguaggio e analisi linguistica: XX incontro di grammatica generativa. Unipress, Padua, pp. 3-23. Alexiadou, Artemis. 1994a. “Adverbial Licensing and Clause Structure’, ms., University of Potsdam. —_——§., 1994b. “Issues in the Syntax of Adverbs”, Ph.D. diss., University of Potsdam.

———. 1997. Adverb Placement. A case study in antisymmetric syntax. Benjamins, Amsterdam. Ameka, Felix K. 1988. “The Grammatical Coding of the Terminal Viewpoint of Situations in Ewe: A Semantic Perspective.” Australian Journal of Linguistics 8:185-217. Anderson, Stephen, and Bernard Comrie, eds. 1991. Tense and Aspect in Eight Languages of Cameroon. Summer Institute of Linguistics and the University of Texas at Arlington. Andrews, Avery. 1982. “A Note on the Constituent Structure of Adverbials and Auxiliaries.” Linguistic Inquiry 13:313-317. —-—-—. 1983. “A Note on the Constituent Structure of Modifiers.” Linguistic Inquiry 14:695-697. Andvik, Erik. 1993. “Tshangla verb inflections.” Linguistics of the Tibeto-Burman Area 16:75—136.

Aqvist, Lennart, Franz Guenther, and Christian Rohrer. 1977. “‘Soon’ and ‘Recently.’” In Christian Rohrer, ed., On the Logical Analysis of Tense and Aspect. Gunter Narr, Tiibingen, pp. 67-81. Arnott, D. W. 1970. The Nominal and Verbal Systems of Fula. Oxford, Clarendon Press. Asher, R. E. and T. C. Kumari. 1997. Malayalam. Routledge, London. Avolonto, Aimé B. 1992a. “AspP et la catégorie INFL en Fongbé.” Journal of West African Languages 22(1):97-113. ———., 1992b. De Il’ étude sémantico-syntaxique des marqueurs pré-verbaux a la structure de la phrase en Fangbe. Mémoire de Maitrise, University of Quebec in Montreal. ———-. 1992c. “Les particules modales en Fongbé et la nature de INFL dans les phrases injonctives.” In Chris Collins and Victor Manfredi, eds., Proceedings of the Kwa Comparative Syntax Workshop, MIT 1992, MIT Working Papers in Linguistics, vol. 17. Cambridge, Mass., pp. 27—40. Aw6bultyi, Oladele. 1967. “Studies in the Syntax of the Standard Yoruba Verb.” Ph.D. diss., Columbia University (distributed by UMI Dissertation Services, Ann Arbor, Michigan). Axelrod, Melissa. 1993. The Semantics of Time: Aspectual Categorization in Koyukon Athapaskan. University of Nebraska Press, Lincoln. Babu, Hany. 1996. “The Binarity of the Malayalam Tense System: Some Preliminary Remarks.” CIEFL Occasional Papers in Linguistics (Hyderabad) 6:1-23. Bache, Carl, Hans Basbgll, and Carl-Erik Lindberg, eds. 1994. Tense Aspect and Action: Empirical and Theoretical Contributions to Language Typology. Mouton de Gruyter, Berlin.

REFERENCES 233

Baker, Carl L. 1971. “Stress Level and Auxiliary Behavior in English.” Linguistic Inquiry 2:167-181. ———., 1981. “Auxiliary-Adverb Word Order.” Linguistic Inquiry 12:309-315. ———., 1989. English Syntax. MIT Press, Cambridge, Mass. Baker, Mark. 1985. “The Mirror Principle and Morphosyntactic Explanation.” Linguistic Inquiry 16:373-415. ———. 1988. Incorporation. University of Chicago Press, Chicago. ———. 1995. The Polysynthesis Parameter. Oxford University Press, New York. Baltin, Mark. 1995. “Floating Quantifiers, PRO, and Predication.” Linguistic Inquiry 26:199-248.

Barnes, Janet. 1994. “Tuyuca.” In Peter Kahrel and René van den Berg, eds., Typological Studies in Negation. John Benjamins, Amsterdam, pp. 325-342. Barr, Donald F. 1988. “Da’a Verbal Affixes and Clitics.” In Hein Steinhauer, ed., Papers in Western Austronesian Linguistics 4:11-—49. Pacific Linguistics, Series A, no. 79. Australian National University, Canberra. Bartsch, Renate. 1976. The Grammar of Adverbials: A Study in the Semantics and Syntax of Adverbial Constructions. North Holland, Amsterdam. Battaglia, Salvatore. 1986. Grande dizionario della lingua italiana. Vol.13. UTET, Turin. Battistella, Edwin L. 1990. Markedness: The Evaluative Superstructure of Language. State University of New York Press, Albany, N.Y. ——. 1991. “The Treatment of Negation in Double Modal Constructions.” Linguistic Analysis 21:49-65. ———. 1995. “The Syntax of the Double Modal Construction.” Linguistica Atlantica 17:19-44. ————.. 1996. The Logic of Markedness. Oxford University Press, New York. Bauer, Winifred. 1993. Maori. Routledge, London.

Bayer, Josef. 1996. Directionality and Logical Form: On the Scope of Focussing Particles and wh-in-situ. Kluwer, Dordrecht. Baxter, Alan N. 1988. A Grammar of Kristang (Malacca Creole Portuguese). Pacific Linguistics, Series B, no. 95. Australian National University, Canberra.

Beaumont, Clive H. 1979. The Tigak Language of New Ireland. Pacific Linguistics, Series B, no. 58. Australian National University, Canberra. Beghelli, Filippo. 1995. “The Phrase Structure of Quantifier Scope.” Ph.D. diss., University of California at Los Angeles. Beghelli, Filippo, and Tim Stowell. 1997. “Distributivity and Negation: The Syntax of each

and every.” In Anna Szabolcsi, ed., Ways of Scope Taking. Kluwer, Dordrecht, pp. 71-107. Bellert, frena. 1977. “On Semantic and Distributional Properties of Sentential Adverbs.” Linguistic Inquiry 8:337-351. Belletti, Adriana. 1990. Generalized Verb Movement. Rosenberg and Sellier, Turin. ——_—. 1994a. “Pas vs. pit: A Note.” In Guglielmo Cinque et al., eds., Paths towards Universal Grammar: Studies in Honor of Richard S. Kayne. Georgetown University Press, Washington, D.C., pp. 25-34. ————. 1994b. “Verb Positions: Evidence from Italian.” In David Lightfoot and Norbert Hornstein, eds., Verb Movement, Cambridge University Press, Cambridge, pp. 19-40. Bellusci, David Christian. 1991. “Serialization Patterns in Shona Verbal Morphology.” M.A. thesis, University of Calgary (distributed by UMI Dissertation Services, Ann Arbor, Michigan). Beninca’, Paola. 1988. “L’ordine degli elementi della frase e le costruzioni marcate.” In Lorenzo Renzi, ed., Grande grammatica italiana di consultazione. Vol. 1. Il Mulino, Bologna, pp. 115-225.

234 REFERENCES

——_—, ed. 1989. Dialect Variation and the Theory of Grammar. Foris, Dordrecht. ——_—., 1994. La variazione sintattica. {1 Mulino, Bologna. ———. 1996. “La struttura della frase esclamativa alla luce del dialetto padovano.” In Paola Beninca’ et al., eds., /taliano e dialetti nel tempo: Saggi di grammatica per Giulio C. Lepschy. Bulzoni, Rome, pp. 23-43.

Beninca’, Paola, and Cecilia Poletto. 1994. “Bisogna and Its Companions: The Verbs of Necessity.” In Cinque et al., eds., Paths towards Universal Grammar: Studies in Honor of Richard S. Kayne. Georgetown University Press, Washington, D.C., pp. 35-57. ——_—., 1996. “The Diachronic Development of a Modal Verb of Necessity.” University of Venice Working Papers in Linguistics 6(2):29-63. Beninca’, Paola, and Laura Vanelli. 1982. ““Appunti di sintassi veneta.” In Manlio Cortelazzo, ed., Guida ai dialetti veneti IV. Cleup, Padua, pp. 7-38.

Bergman, Brita, and Osten Dahl. 1994. “Ideophones in Sign Language? The Place of Reduplication in the Tense-Aspect System of Swedish Sign Language.” In Carl Bache et al., eds., Tense Aspect and Action: Empirical and Theoretical Contributions to Language Typology. Mouton de Gruyter, Berlin, pp. 397-422. Bergsland, Knut. 1994. “Aleut tenses and aspects.” In Carl Bache et al., eds., Tense Aspect and Action: Empirical and Theoretical Contributions to Language Typology, Mouton de Gruyter, Berlin, pp. 323-369.

Bernstein, Michael. 1994. “Some Notes on Infinitives and Negation in English.” Cornell Working Papers in Linguistics 12:1-25. Bertinetto, Piermarco. 1994a. “Le perifrasi abituali in italiano e in inglese.” Unpublished ms., Scuola Normale Superiore, Pisa. —_—_—-, 1994b. “Statives, Progressives and Habituals: Analogies and Divergencies.” Linguis-

tics 32:391-423. ———., 1994c. “Temporal Reference, Aspect and Actionality: Their Neutralization and Interactions, Mostly Exemplified in Italian.” In Carl Bache et al., eds., Tense Aspect and Action: Empirical and Theoretical Contributions to Language Typology. Mouton de Gruyter, Berlin, pp. 113-137. Bhat, D. N. S. 1968. Boro Vocabulary (with a grammatical sketch). Deccan College, Pune. —___—, 1994, The Adjectival Category. John Benjamins, Amsterdam. Bhatt, Rakesh, and James Yoon. 1991. “On the Composition of COMP and Parameters of V2.” Proceedings of the 10th West Coast Conference on Formal Linguistics, Stanford Linguistic Association, Stanford, Calif., pp. 41-52. Bickerton, Derek. 1974. “Creolization, Linguistic Universals, Natural Semantax and the Brain.” University of Hawaii Working Papers in Linguistics 6(3):124—141. —___—, 1984. “The Language Bioprogram Hypothesis.” Behavioral and Brain Sciences 7 (2):173-221. Binnick, Robert I. 1991. Time and the Verb: A Guide to Tense and Aspect. Oxford University Press, New York. Blansitt, Edward L. Jr. 1975. “Progressive Aspect.” Working Papers on Language Universals 18:1—34 (Stanford University). Bliese, Loren F. 1981. A Generative Grammar of Afar. Summer Institute of Linguistics and the University of Texas at Arlington. Bobalijk, Jonathan D., and Dianne Jonas. 1996. “Subject Positions and the Roles of TP.” Linguistic Inquiry 27:195—236. Bodomo, Adams B. 1993. “Dagaare Syntax: A Two-Level X-Bar Account.” University of Trondheim Working Papers in Linguistics 19:20—51. Boertien, Harmon S. 1986. “Constituent Structure of Double Modals.” In Michael B. Montgomery and Guy Bailey, eds., Language Variety in the South. University of Alabama Press, Tuscaloosa, pp. 294-318.

REFERENCES 235

Borer, Hagit. 1996. “Passive without Theta Grids.” In Steve Lapointe, ed., Morphological Interfaces. CSLI, Stanford University. Borgato, Gianluigi., ed. 1994. Teoria del linguaggio e analisi linguistica: XX tncontro di grammatica generativa. Unipress, Padua. Botne, Robert. 1993. “Differentiating the Auxiliaries -Ti and -Va in Tumbuka (N.21).” Linguistique Africaine 10:7-27. Boutin, Michael E. 1991. “Aspect and Temporal Reference in Banggi.” In Stephen H. Levinson, ed., Thematic Continuity and Development in Languages of Sabah. Pacific Linguistics, Series C, no.118. Australian National University, Canberra, pp. 7~28. Bowers, John. 1993. “The Syntax of Predication.” Linguistic Inquiry 24:591-656. Brennan, Virginia Mary. 1993. “Root and Epistemic Modal Auxiliary Verbs.” Ph.D. diss., University of Massachusetts at Amherst. Brody, Michael. 1990. “Remarks on the Order of Elements in the Hungarian Focus Field.” In Istvan Kenesei, ed., Approaches to Hungarian. Vol. 3. Jate, Szeged, pp. 95-121. Brown, Keith. 1992. “Double Modals in Hawick Scots.” In Peter Trudgill and J. K. Chambers, eds., Dialects of English: Studies in Grammatical Variation. Longman, London, pp. 74-103. Brown, Penelope, and Stephen Levinson. 1987. Politeness: Some Universals in Language Usage. Cambridge University Press, Cambridge. Browning, Maggie. 1996. “CP Recursion and that-t Effects.” Linguistic Inquiry 27:237-255. Brugger, Gerhard, and Cecilia Poletto. 1993. “On Negation in German and Bavarian.” University of Venice Working Papers in Linguistics 3 (2):41-79. Brugger, Gerhard, and Mario D’ Angelo. 1995. “Movement at LF Triggered by Mood and Tense.” Folia Linguistica 29:195—221. Burzio, Luigi. 1981, “Intransitive Verbs and Italian Auxiliaries.” Ph.D. diss., MIT, Cambridge, Mass. ———, 1986. Italian Syntax. Kluwer, Dordrecht. Bybee, Joan. 1985. Morphology: A Study of the Relation between Meaning and Form. John Benjamins, Amsterdam. Bybee, Joan, and Suzanne Fleischman, eds. 1995. Modality in Grammar and Discourse. John Benjamins, Amsterdam. Bybee, Joan, Revere Perkins, and William Pagliuca. 1994. The Evolution of Grammar: Tense, Aspect and Modality in the Languages of the World. University of Chicago Press, Chicago. Calabrese, Andrea. 1984. “Multiple Questions and Focus in [talian.” In Wim de Geest and Yvan Putseys, eds., Sentential Complementation. Foris, Dordrecht, pp. 67-74. ————, 1985. “The Mirror Principle and the Latin Passive.” Ms., MIT.

Campbell, George L. 1991. Compendium of the World’s Languages. Routledge, London. Campbell, Richard. 1996. “Serial Verbs and Shared Arguments.” Linguistic Review 13:83~-118.

Camporese, Nadia. 1996. “Gli avverbi e la struttura del predicato in cinese.” Tesi di laurea, University of Venice. Capell, A., and H. H. J. Coate. 1984. Comparative Studies in Northern Kimberley Languages. Pacific Linguistics, Series C, no. 69. Australian National University, Canberra. Cardinaletti, Anna. 1991. “On Pronoun Movement: The Italian Dative Loro.” Probus 3:127—-153.

———. 1997, “Subjects and Clause Structure.” In Liliane Haegeman, ed., The New Comparative Syntax. Longman, London, pp. 33-63. Cardinaletti, Anna, and Maria Teresa Guasti. 1993. “Negation in Epistemic Small Clauses.” Probus 5:39-61. —._-, eds. 1995. Small Clauses. Academic Press, New York. Cardinaletti, Anna, and Michal Starke. 1994. “The Typology of Structural Deficiency: On the Three Grammatical Classes.” University of Venice Working Papers in Linguistics 4 (2):41—-109.

236 REFERENCES

Carlson, Barry. 1996. “Situation Aspect and a Spokane Control Morpheme.” International Journal of American Linguistics 62:59-69. Chafe, Wallace. 1986. “Evidentiality in English Conversation and Academic Writing.” In Wallace Chafe and Johanna Nichols, eds., Evidentiality: The Linguistic Coding of Epistemology. Ablex, Norwood, N.J., pp. 261-272. Chafe, Wallace, and Johanna Nichols, eds. 1986. Evidentiality: The Linguistic Coding of Epistemology. Ablex, Norwood, N.J. Chhangte, Lalnunthangi. 1986. “A Preliminary Grammar of the Mizo Language.” Master’s thesis, University of Texas at Arlington (distributed by UMI Dissertation Services, Ann Arbor, Michigan). ————., 1989. “The Grammar of Simple Clauses in Mizo”, in David Bradley, ed. Papers in South-East Asian Linguistics 11.93-174. Pacific Linguistics, Series A, no. 77. Australian National University, Canberra. Chia, Emmanuel N. 1976. “Kom Tenses and Aspects.” Ph.D. diss., Georgetown University. Chierchia, Gennaro. Forthcoming. “Individual Level Predicates as Inherent Generics.” In Greg Carlson and F. J. Pelletier, eds., The Generic Book. University of Chicago Press, Chicago. Cho, Dong-In. 1994. “Functional Projections and Verb Movement.” In Young-Key KimRenaud, ed., Theoretical Issues in Korean Linguistics. Stanford Linguistics Society, Stanford University, pp. 233—254. Chomsky, Noam. 1957. Syntactic Structures. Mouton, The Hague. ——_———. 1965. Aspects of the Theory of Syntax. MIT Press, Cambridge, Mass. ————. 1970. “Remarks on Nominalization.” In Roderick A. Jacobs and Peter S. Rosenbaum, eds., Readings in English Transformational Grammar. Ginn, Waltham, Mass., pp. 184-221. —————. 1986. Barriers. MIT Press, Cambridge, Mass.

————. 1995. The Minimalist Program. MIT Press, Cambridge, Mass. Chung, Sandra, and Alan Timberlake. 1985, “Tense, Aspect and Mood.” In Timothy Shopen, ed., Language Typology and Syntactic Description, Vol. 3. Cambridge University Press, Cambridge, pp. 202~258. Cinque, Guglielmo. 1976. “Mica.” In Annali della facolta di lettere e filosofia dell’ Universita

di Padova, pp. 101-112. ——_———., 1983. “Su una differenza tra italiano e inglese nelle costruzioni ‘ad ellissi dell’ oggetto’.” Rivista di Grammatica Generativa 8:127-151. ——-—-, 1990a. “Ergative Adjectives and the Lexicalist Hypothesis.” Natural Language and Linguistic Theory 8:1—40. ——————, 1990b. Types of A’-Dependencies. MIT Press, Cambridge, Mass.

—_——, 1991. Teoria linguistica e sintassi italiana. 11 Mulino, Bologna. —__——, 1992a. “On Leftward Movement of tutto in Italian.” University of Venice Working Papers in Linguistics 2:6. ————, 1992b. “On tout/tutto and the Syntax of Past Participles in French and Italian.” Unpublished ms., University of Venice. —_——. 1993. “A Null Theory of Phrase and Compound Stress.” Linguistic Inquiry 24:239297. ————-. 1994. “Evidence for Partial N-movement in the Romance DP.” In Guglielmo Cinque et al., eds., Paths towards Universal Grammar: Studies in Honor of Richard S. Kayne. Georgetown University Press, Washington, D.C., pp. 85-110. ——_—— ., 1995. Italian Syntax and Universal Grammar. Cambridge University Press, Cambridge.

—_—— . 1996. “The Antisymmetric Programme: Theoretical and Typological Implications.”

Journal of Linguistics 32:447-464. Cinque, Guglielmo, et al., eds. 1994. Paths towards Universal Grammar: Studies in Honor of Richard S. Kayne. Georgetown University Press, Washington, D.C.

REFERENCES 237

Cloarec, France. 1972. Le verbe banda. SELAF, Paris. Cole, Peter. 1985. Imbabura Quechua. Croom Helm, London. Comrie, Bernard. 1976a. Aspect. Cambridge University Press, Cambridge. ——. 1976b. “Linguistic Politeness Axes: Speaker-Addressee, Speaker-Referent, SpeakerBystander.” Pragmatics Microfiche 1(7):1-12. ———, 1981. “On Reichenbach’s Approach to Tense.” In Papers from the 17th Regional Meeting of the Chicago Linguistic Society. Chicago Linguistic Society, Chicago, pp. 24—30. ———, 1985. Tense. Cambridge University Press, Cambridge. Corne, Chris. 1977. Seychelles Creole Grammar. Gunter Narr, Tiibingen. Corum, Claudia. 1974. “Adverbs ... Long and Tangled Roots.” In Papers from the 10th Regional Meeting of the Chicago Linguistic Society. Chicago Linguistic Society, Chicago, pp. 90-102. Corver, Norbert. 1997a. “The Internal Syntax of the Dutch Extended Adjectival Projection.” Natural Language and Linguistic Theory 15:289-368. ——. 1997b. “Much-Support as a Last Resort.” Linguistic Inquiry 28:119-164. Corver, Norbert, and Henk van Riemsdijk, eds. 1994. Studies on Scrambling. Mouton de Gruyter, Berlin. Costa, Jofio. 1996. “Adverb Positioning and V-Movement in English: Some More Evidence.” Studia Linguistica 50:22—-34.

Cram, David. 1983/84. “Scottish Gaelic ANN as an Aspect Particle.” Studia Celtica 18/ 19:311-326. Cresswell, M. J. 1985. Adverbial Modification. Reidel, Dordrecht. Cunningham, Irma A. E. 1992. A Syntactic Analysis of Sea Island Creole. University of Alabama Press, Tuscaloosa. Cuvalay, Martine. 1995, “The E-Structure in Functional Grammar: Towards a Consistent Treatment of Tense, Mood, Aspect and Illocutionary Force.” Working Papers in Functional Grammar, no. 59. IFOTT, University of Amsterdam. Dahl, Osten. 1979. “Typology of Sentence Negation.” Linguistics 17:79-106.

——. 1983. “Temporal Distance: Remoteness Distinctions in Tense-Aspect Systems.” Linguistics 21:105—122. ——_——, 1985. Tense and Aspect Systems. Blackwell, Oxford. Dahl, Osten, and Dora K6s-Dienes, eds. 1984. Selected Working Papers from the Tense-Mood-

Aspect Project, Institute of Linguistics. University of Stockholm. Daoust-Blais, Denise, and William Kemp. 1979. “Pour pas tout que ca se perde: pour as a ‘quantifier raising’ subordinator in Quebec French.” Unpublished ms., University of Montreal. Davidson, Donald. 1967. “The Logical Form of Action Sentences.” In Nicholas Rescher, ed., The Logic of Decision and Action. University of Pittsburgh Press, Pittsburgh, pp. 81-95. Davies, H. J. 1981. “The Syntax of the Simple Sentence in Kobon.” Papers in New Guinea Linguistics 21:1—70. Pacific Linguistics, Series A, no. 61. Australian National University, Canberra. Déchaine, Rose-Marie. 1991. “Bare Sentences.” Cornell Working Papers in Linguistics 10:3 1—S0.

—_—_—. 1993. “Predicates Across Categories.” Ph.D. diss., University of Massachusetts, Amherst. DeGraff, Michel A. F, 1993. “Creole Grammars and the Acquisition of Syntax: The Case of Haitian.” Rev. ed. Ph.D. diss., University of Pennsylvania. ——., 1994. “To Move or Not to Move? Placement of Verbs and Object Pronouns mm Haitian Creole and in French.” In Papers from the 30" Regional Meeting of the Chicago Linguistic Society, Chicago Linguistic Society, Chicago, pp. 141-155.

238 REFERENCES

———, 1996. “Review Article of Temps et aspects en créole Seychellois: Valeurs et interférences, by Susanne Michaelis. Hamburg: Buske, 1993.” Journal of Pidgin and Creole Languages 11:121-137. de Groot, Casper, 1984. “Totally Affected: Aspect and Three-Place Predicates in Hungarian.” In Casper de Groot and Hannu Tommola, eds., Aspect Bound: A Voyage into the

Realm of Germanic, Slavonic and Finno-Ugrian Aspectology. Foris, Dordrecht, pp. 133-151. ———_—. 1995. “Aspect, Mood and Tense in Functional Grammar.” In Pier Marco Bertinetto

et al., eds., Typological Perspectives. Vol. 2 of Temporal Reference, Aspect and Actionality. Rosenberg and Sellier, Turin, pp. 29-45. de Groot, Casper, and Hannu Tommola., eds. 1984. Aspect Bound: A Voyage into the Realm of Germanic, Slavonic and Finno-Ugrian Aspectology. Foris, Dordrecht. Delancey, Scott. 1986. “Evidentiality and Volitionality in Tibetan.” In Wallace Chafe and Johanna Nichols, eds., Evidentiality: The Linguistic Coding of Epistemology. Ablex, Norwood, N.J., pp. 203-213. —— . 1997. “Mirativity: The Grammatical Marking of Unexpected Information.” Linguistic Typology 1:33-52. Delfitto, Denis, and Piermarco Bertinetto. 1994. “A Case Study in the Interaction of Aspect and Actionality: The Imperfect in Italian.” In Piermarco Bertinetto et al., eds., Temporal Reference, Aspect and Actionality, vol. 1. Rosenberg and Sellier, Turin, pp. 125-142. Dendale, Patrick, and Liliane Tasmowsky, eds. 1994. “Les sources du savoir et leur marques linguistiques.” Langue francaise, n. 102. Déprez, Viviane. Forthcoming. “The Roots of Negative Concord in French and French Based Creoles.” In Michel DeGraff, ed., Parametric Approaches to Creole Syntax. MIT Press, Cambridge, Mass. Déprez, Viviane, and Marie-Therése Vinet. 1992. “Une structure prédicative sans copule.” Revue Québécoise de Linguistique 22:11—44. Derbyshire, Desmond C. 1985. Hixkaryana and Linguistic Typology. Summer Institute of Linguistics and the University of Texas at Arlington. —-—_—. 1986. “Comparative Survey of Morphology and Syntax in Brazilian Arawakan.” In Desmond C. Derbyshire and Geoffrey K. Pullum, eds., Handbook of Amazonian Languages, vol.1. Mouton de Gruyter, Berlin, pp. 469-566. Diesing, Molly. 1992. Indefinites. MIT Press, Cambridge, Mass. ———, 1996a. “NP Types and Conditions on Interpretations.” In OTS Yearbook 1995. OTS, Utrecht University, pp. 1-12. ———., 1996b. “Semantic Variables and Object Shift.” In Héskuldur Thrainsson, Samuel D. Epstein, and Steve Peter, eds., Studies in Comparative Germanic Syntax, Vol. 2. Kluwer, Dordrecht, pp. 66-84. Diesing, Molly, and Eloise Jelinek. 1995. “Distributing Arguments.” Natural Language Semantics 3:123—176.

Dik, Simon. 1975. “The Semantic Representation of Manner Adverbials.” In A. Kraak, ed., Linguistics in the Netherlands 1972-1973. Van Gorcum, Assen/Amsterdam, pp. 96-121. ————, 1989. The Theory of Functional Grammar. Part I: The Structure of the Clause. Foris, Dordrecht.

Dik, Simon, Kees Hengeveld, Elseline Vester, and Co Vet. 1990. “The Hierarchical Structure of the Clause and the Typology of Adverbial Satellites.” In Nuyts et al., eds., Layers and Levels of Representation in Language Theory. John Benjamins, Amsterdam, pp. 25-— 70. Dixon, Robert M. W. 1972. The Dyrbal Language of North Queensland. Cambridge University Press, Cambridge.

REFERENCES 239

———, ed. 1976. Grammatical Categories in Australian Languages. Australian Institute of Aboriginal Studies, Canberra. Dobrovie-Sorin, Carmen. 1990. “Auxiliaries and Sentence Structure in Rumanian.” In Joan Mascaro and Marina Nespor, eds., Grammar in Progress. GLOW Essays for Henk van Riemsdijk. Foris, Dordrecht. —_——. 1994. The Syntax of Romanian. Mouton de Gruyter, Berlin. Dobrovolsky, Michael. 1976. “Is Turkish an Agglutinating Language?” In Papers from the 6th Meeting of the North Eastern Linguistic Society. L’ association linguistique de Montréal, Montréal, pp. 87-101. Doherty, Monika. 1985. Epistemische Bedeutung. Studia Grammatica, no. 23. AkademieVerlag, Berlin. ———.. 1987. “Perhaps.” Folia Linguistica 21:45-—65. Donaldson,Tamsin. 1980. Ngiyambaa. Cambridge University Press, Cambridge. Dowty, David. 1979. Word Meaning and Montague Grammar. Reidel, Dordrecht. Drijkoningen, Frank. 1986. “On the Order of (Auxiliary) Verbs and Syntactic Affixes.” In Frits Beukema and Aafke Hulk, eds., Linguistics in the Netherlands 1986, Dordrecht, Foris, pp. 51-59. Dryer, Matthew S. 1988. “Universals of Negative Position.” In Michael Hammond, Edith Moravesik, and Jessica Wirth, eds., Studies in Syntactic Typology. John Benjamins, Amsterdam, pp. 93-124. Dubuisson, Claire, and John Goldsmith. 1976. “A propos de l’inversion du clitique sujet en

francais.” Papers from the 6th Meeting of the North Eastern Linguistic Society. L’ association linguistique de Montréal, Montréal, pp. 103-112. Du Feu, Veronica. 1996. Rapanui. Routledge, London. Dugast, Idelette. 1971. Grammaire du Tinen. Klincksieck, Paris. Karly, Robert. 1994. “Lewo.” In Peter Kahrel and René van den Berg, eds., Typological Studies in Negation. John Benjamins, Amsterdam, pp. 65-972. Egerland, Verner. 1996. “On Pronoun Positions in Swedish and Italian, Antisymmetry, and the Person Phrase.” University of Venice Working Papers in Linguistics 6 (2):65—104.

Eng, Miirvet. 1987, “Anchoring Conditions for Tense.” Linguistic Inquiry 18:633-657. England, Nora C. 1983. A Grammar of Mam, a Mayan Language. University of Texas Press, Austin. Engver, Karl. 1972. Place de l’adverbe déterminant un infinitif dans la prose du francais contemporain. Studia Romanica Upsaliensia, 7, Uppsala (distributed by Almqvist & Wiksell, Stockholm).

Ernst, Thomas B. 1984. Towards an Integrated Theory of Adverb Position in English. Bloomington, Indiana University Linguistics Club. ———. 1987. “Why Epistemic and Manner Modifications Are Exceptional.” In Proceedings of the 13th Annual Meeting of the Berkeley Linguistics Society, Linguistics Society, Berkeley, Calif., pp. 77—87. ———. 1991. “On the Scope Principle.” Linguistic Inquiry 22:750-756. ————-. 1992. “The Phrase Structure of English Negation.” Linguistic Review 9:109-144. Ernst, Urs. 1991. “Temps et aspect en kako.” In Stephen Anderson and Bernard Comrie, eds., Tense and Aspect in Eight Languages of Cameroon. Summer Institute of Linguistics and the University of Texas at Arlington, pp. 17-45. Erteshik-Shir, Nomi. 1973. “On the Nature of Island Constraints.” Ph.D. diss., MIT. Escalante, Fernando. 1984. “Moods and Modes in Yaqui.” In Stuart Davis, ed., Studies on Native American Languages, Japanese and Spanish. Coyote Papers. Working Papers in Linguistics from A to Z. University of Arizona, Tucson, vol. 5, pp. 19~31.

240 REFERENCES

Espinal, Maria Teresa. 1984. “Analisi interpretativa d’encara i ja.” Estudis gramaticals I. Working Papers in Linguistics. Universitat Autonoma de Barcelona, pp. 109-148. ———. 1991. “Negation in Catalan. Some Remarks with Regard to no pas.” Catalan Working Papers in Linguistics 1991. Universitat Autonoma de Barcelona, pp. 33-63. —_——., 1993. “The Interpretation of no-pas in Catalan.” Journal of Pragmatics 19:353-369.

Essien, Okon E. 1987. “The Aspectual System of Ibibio.” In David Odden, ed., Current Approaches to African Linguistics, vol. 4. Foris, Dordrecht, pp. 151-165. Evans, Nicholas D. 1995. A Grammar of Kayardild. Mouton de Gruyter, Berlin. Fagerli, Ole Torfinn. 1994. “Verbal Derivations in Fulfulde.” Cand. Philol. diss., University of Trondheim. Fassi Fehri, Abdelkader. 1993. Issues in the Structure of Arabic Clauses and Words. Kluwer, Dordrecht. ———., 1997. “Arabic Adverbs: A Preliminary Investigation.” IERA, Rabat. ————. Forthcoming. “Layers in the Distribution of Arabic Adverbs and Adjectives and Their Licensing.” In M. Eid, ed., Perspectives on Arabic Linguistics 11. John Benjamins, Amsterdam. Fillmore, Charles J. 1994. “Under the Circumstances (Place, Time, Manner, etc.).” In Proceedings of the 20th Annual Meeting of the Berkeley Linguistics Society, Linguistics Society, Berkeley, Calif., pp. 158-172. Foley, William A. 1986. The Papuan Languages of New Guinea. Cambridge University Press, Cambridge. ———. 1991. The Yimas Language of New Guinea. Stanford University Press, Stanford, Calif. Foley, William A., and Robert D. Van Valin. 1984. Functional Syntax and Universal Grammar. Cambridge University Press, Cambridge.

Forrest, Tracey. 1990. “The Use of AP and the VA/AP Transition in Haitian Creole.” CUNYForum 15:19-25. Fortescue, Michael D. 1980. “Affix Ordering in West Greenlandic Derivational Processes.” International Journal of American Linguistics 46:259-278. ———., 1984. West Greenlandic. Croom Helm, London. Fox, Greg J. 1979. Big Nambas Grammar. Pacific Linguistics, Series B, no. 60. Australian National University, Canberra. Frank, Paul. 1990. Ika Syntax. University of Texas Press, Arlington. Fraser, Bruce. 1996. “Pragmatic Markers.” Pragmatics 6:167—-190. Frawley, William. 1992. Linguistic Semantics. Lawrence Erlbaum, Hillsdale, N_J.

Freed, Alice F. 1979. The Semantics of English Aspectual Complementation. Reidel, Dordrecht. Frick, Esther. 1978. “The Verbal System in Dghwede.” Linguistics 212:5-43. Friedman, Victor. 1986. “Evidentiality in the Balkans: Bulgarian, Macedonian and Albanian.” In Wallace Chafe and Johanna Nichols, eds., Evidentiality: The Linguistic Coding of Epistemology. Ablex, Norwood, N.J., pp. 168-187. Gamble, Geoffrey. 1978. Wikchamni Grammar. University of California Press, Berkeley. Gibson, Kean. 1986. “The Ordering of Auxiliary Notions in Guyanese Creole.” Language 62:571-586. ——-—, 1992, “Tense and Aspect in Guyanese Creole with Reference to Jamaican and Carriacouan.” International Journal of American Linguistics 58:49-95. Giorgi, Alessandra, and Fabio Pianesi 1991. “Toward a Syntax of Temporal Representations.” Probus 3:187-213. ———, 1996. “Verb Movement in Italian and Syncretic Categories.” Probus 8:137~-160. _—~--, 1997, Tense and Aspect: From Semantics to Morphosyntax. Oxford University Press, New York.

REFERENCES 241

Giusti, Giuliana. 1991. “The Categorial Status of Quantified Nominals.” Linguistische Berichte 136:438-454. ———~, 1993. La sintassi dei determinanti. Unipress, Padua. Givon, Talmy. 1972. Studies in ChiBemba and Bantu Grammar. Studies in African Linguistics, Supplement 3. ———. 1982a. “Evidentiality and Epistemic Space.” Studies in Language 6:23-49. ———-. 1982b. “Tense-Aspect-Modality: The Creole Prototype and Beyond.” In Paul Hopper, ed., Tense-Aspect: Between Semantics and Pragmatics. John Benjamins, Amsterdam,

pp. 115-163. Goddard, Ives (ed.) 1996. Languages (vol. 17 of Handbook of North American Indians). Smithsonian Institution, Washington. Goral, Donald R. 1988. “Modal Verbs and Generative Grammar.” Linguistics of the TibetoBurman Area 1\1:7-24. Greenberg, Joseph H. 1966. Language Universals: With Special Reference to Feature Hierarchies. Mouton, The Hague. Grimshaw, Jane. 1991. “Extended Projections.” Unpublished ms., Brandeis University, Waltham, Mass. Griset, Ilia. 1966. La parlata provenzaleggiante di Inverso Pinasca. Giappichelli Editore, Turin. Grohmann, Kleanthes Kostas. 1997. “Pronouns and the left periphery of West Germanic embedded clauses”, in Werner Abraham and Elly van Gelderen, eds., German: Syntactic Problems-Problematic Syntax. Niemeyer, Tiibingen, pp. 163-189. Groves, Terab’ata R., Gordon W. Groves, and Roderick Jacobs. 1985. Kiribatese: An Outline Description. Pacific Linguistics, Series D, no. 64. Australian National University, Canberra. Haegeman, Liliane. 1993. “Some Speculations on Argument Shift, Clitics and Crossing in West Flemish.” In Werner Abraham and Josef Bayer, hrsg., Dialektsyntax. Linguistische Berichte, Sonderheft 5. Westdeutscher Verlag, Opladen, pp. 131-160. Haftka, Brigitta. 1995. “Syntactic Positions for Topic and Contrastive Focus in the German Middlefield.” In Inga Kohlhof, Susanne Winkler, and Hans Bernhard Drubig, eds., Proceedings of the Géttingen Focus Workshop. Bericht no. 69, Sonderforschungsbereichs 340. University of Stuttgart, pp. 137-157. Haiden, Martin. 1995. “On the Semantic Evaluation of NP-Syntax.” Diplomarbeit, University of Vienna. —_——. 1996. “The Aspect of Short Scrambling.” Wiener Linguistische Gazette 37-59:121-145.

Haider, Hubert, and Martin Prinzhorn, eds. 1986. Verb Second Phenomena in Germanic Languages. Foris, Dordrecht. Haider, Hubert, Susan Olsen, and Sten Vikner, eds. 1995. Studies in Comparative Germanic Syntax. Kluwer, Dordrecht. Haiman, John. 1980. HUA: A Papuan Language of the Eastern Highlands of New Guinea. John Benjamins, Amsterdam. Harris, James W. 1991. “The Exponence of Gender in Spanish.” Linguistic Inquiry 22:27-62. Haspelmath, Martin. 1993. A Grammar of Lezgian. Mouton de Gruyter, Berlin. Heath, Daniel. 1991. “Tense and Aspect in Makaa.” In Stephen Anderson and Bernard Comrie, eds., Tense and Aspect in Eight Languages of Cameroon. Summer Institute of Linguistics and the University of Texas at Arlington, pp. 3-15. Heine, Bernd. 1992. “Grammaticalization Chains.” Studies in Language 16:335-368. ———. 1994. “On the Genesis of Aspect in African Languages: The Proximative.” Proceedings of the 20th Annual Meeting of the Berkeley Linguistics Society. Linguistics Soci-

ety, Berkeley, Calif., pp. 35-46.

242 REFERENCES

————. 1995. “Agent-Oriented vs. Epistemic Modality: Some Observations on German Modals.” In Joan Bybee and Suzanne Fleischman, eds., Modality in Grammar and Discourse. John Benjamins, Amsterdam, pp. 15-53. Hendrick, Randall. 1991. “The Morphosyntax of Aspect.” Lingua 85:171-210. Hengeveld, Kees. 1989. “Layers and Operators in Functional Grammar.” Journal of Linguistics 25:127-157.

———. 1990. “The Hierarchical Structure of Utterances.” In Jan A. Nuyts, Machtelt Bolkenstein, and Co Vet, eds., Layers and Levels of Representation in Language Theory. John Benjamins, Amsterdam, pp. 1-23. ————. Forthcoming a. “Adverbs in Functional Grammar.” In Gera Wotiak, ed., Functional Lexicology. ————. Forthcoming b. “Mood and Modality.” In Geert Booij, Christian Lehmann, and Joachim Mugdan, eds., Morphology: A Handbook on Inflection and Word Formation. Mouton de Gruyter, Berlin. Heny, Frank, and Barry Richards, eds. 1983. Linguistic Categories: Auxiliaries and Related Puzzles. Reidel, Dordrecht. Hercus, L.A. 1982. The Bagandji Language. Pacific Linguistics, Series B, no. 67. Australian National University, Canberra. Herring, Susan C. 1988. “Aspect as a Discourse Category in Tamil.” Proceedings of the 14th

Annual Meeting of the Berkeley Linguistics Society. Linguistics Society, Berkeley, Calif., pp. 280-292. Hetland, Jorunn. 1992. Satzadverbien im Fokus. Gunter Narr, Ttibingen. Hewitt, B. G. 1979. “The Expression of ‘Inferentiality’ in Abkhaz.” Journal of Linguistics 15:87-92. ———. 1989. Abkhaz. Routledge, London. Hill, Jane H. 1969. “Volitional and Non-Volitional Verbs in Cupeno.” In Papers from the Sth Regional Meeting of the Chicago Linguistic Society. Chicago Linguistic Society, Chi-

cago, pp. 348-356. Hoekstra, Eric. 1992. “On the Parametrization of Functional Projections in CP.” In Proceedings of the 22nd North Eastern Linguistic Society: 191-204. ———. 1993. “Dialectal Variation inside CP as Parametric Variation.” In Werner Abraham and Josef Bayer, hbrsgg., Dialektsyntax. Linguistische Berichte, Sonderheft 5. Westdeutscher Verlag, Opiaden pp. 161-179. Hoff, B. J. 1986. “Evidentiality in Carib.” Lingua 69:49-103. Holisky, Dee Ann. 1987. “The Case of the Intransitive Subject in Tsova Tush (Batsbi).” Lingua

71:103-132. Hollingsworth, Kenneth R. 1991. “Tense and Aspect in Mofu-Gudur.” In Stephen Anderson and Bernard Comrie, eds., Tense and Aspect in Eight Languages of Cameroon. Summer Institute of Linguistics and the University of Texas at Arlington, pp. 239-255. Holmberg, Anders. 1986. “Word Order and Syntactic Features in the Scandinavian Languages and English.” Ph.D. diss., University of Stockholm. ———. 1993. “Two Subject Positions in IP in Mainland Scandinavian.” Working Papers in Scandinavian Syntax 52:29-41. Holmberg, Anders, et al. 1993. “The Structure of INFL and the Finite Clause in Finnish.” In

Anders Holmberg and Urpo Nikanne, eds., Case and Other Functional Categories in Finnish Syntax. Mouton de Gruyter, Berlin, pp. 177-206. Holmberg, Anders, and Christer Platzack. 1995. The Role of Inflection in Scandinavian Syntax. Oxford University Press, Oxford. Hopper, Paul J., ed. 1982. Tense-Aspect: Between Semantics and Pragmatics. John Benjamins, Amsterdam.

REFERENCES 243

Hornstein, Norbert. 1977. “Towards a Theory of Tense.” Linguistic Inquiry 8:521-557. —————, 1990. As Time Goes By: Tense and Universal Grammar. MIT Press, Cambridge, Mass.

Huang, Shuan-Fan. 1975. A Study of Adverbs. Mouton, The Hague. Hudson, Joyce. 1976. “Walmadyjari.” In Robert M. W. Dixon, ed., Grammatical Categories

in Australian Languages. Australian Institute of Aboriginal Studies, Canberra, pp. 653-667. Huttar, George L., and Mary L. Huttar. 1994. Ndyuka. Routledge, London. Hyman, Larry M. 1980. “Relative Time Reference in the Bamileke Tense System.” Studies in African Linguistics 11:227-—237. Iatridou, Sabine. 1990. “The Past, the Possible, and the Evident.” Linguistic Inquiry 21:123129. Ikegami, Yoshihiko. 1985. “‘Activity’—‘Accomplishment’—‘Achievement’—A Language That Can’t Say I Burned It, but It Didn’t Burn and One That Can.” In Adam Makkai and Alan

K. Melby, eds., Linguistics and Philosophy: Essays in Honor of Rulon S. Wells. John Benjamins, Amsterdam, pp. 265-304. Ikoro, Suanu. 1996. The Kana Language. Research School CNWS, Leiden. Inman, Michael. 1992. “Intentionality in Sinhala.” In Papers from the 28th Regional Meeting of the Chicago Linguistic Society. Chicago Linguistic Society, Chicago, pp. 239-250. Jackendoff, Ray. 1972. Semantic Interpretation in Generative Grammar. MIT Press, Cambridge, Mass. —_———. 1973. “The Base Rules for Prepositional Phrases.” In Stephen Anderson and Paul Kiparsky, eds., A Festschrift for Morris Halle, Holt, Rinehart and Winston, New York, pp. 345-356. ————. 1977. X-Bar Syntax: A Study of Phrase Structure. MIT Press, Cambridge, Mass. ————. 1990. Semantic Structures. MIT Press, Cambridge, Mass. Jackson, Ellen M. 1980. “Aspect, Tense and Time Shifts in Tikar.” Journal of African Languages and Linguistics 2:17-37. Jacobsen, William H. 1986. “The Heterogeneity of Evidentials in Makah.” In Wallace Chafe and Johanna Nichols, eds., Evidentiality: The Linguistic Coding of Epistemology. Ablex, Norwood, N.J., pp. 3-28. Jakobson, Roman. 1939/1971. “Signe zéro.” In Roman Jakobson, Selected Writings, vol. 2. Mouton, The Hague, pp. 211-219. ————, 1957/1971. “Shifters, Verbal Categories, and the Russian Verb.” In Roman Jakobson,

Selected Writings, vol. 2. Mouton, The Hague, pp. 130-147. Jensen, Allen. 1994. “Wayampi.” In Peter Kahrel and René van den Berg, eds., Typological Studies in Negation. John Benjamins, Amsterdam, pp. 343-364. Jensen, John Thayer. 1977. Yapese Reference Grammar. University of Hawaii Press, Honolulu. Jespersen, Otto. 1924. The Philosophy of Grammar. Allen & Unwin, London. Jonas, Dianne. 1996. “Clause Structure, Expletives and Verb Movement.” In Werner Abraham

et al., eds., Minimal Ideas. Syntactic Studies in the Minimalist Framework. John Benjamins, Amsterdam, pp. 167~188. Jones, Frederick C. V. 1990. “The Grammatical Items bin, fo and mos in Sierra Leone Krio.” Linguistics 28:845—866. Jones, Michael Alan. 1993. Sardinian Syntax. Routledge, London. Jones, Robert B. J. 1961. Karen Linguistic Studies. University of California Press, Berkeley. Kahrel, Peter, and René van den Berg, eds. 1994. Typological Studies in Negation. John Benjamins, Amsterdam. Kallulli, Dalina. 1995. “Clitics in Albanian.” Cand. Philol. diss., University of Trondheim.

Kamarah, Sheikh Umarr. 1997. “Tense and Aspect in Ka-Themne,” African Journal of Languages and Linguistics, 2(1).23—40.

244 REFERENCES

Kana, Marit. 1983. “The Syntax and Semantics of Verbal Prefixes in Bahasa Indonesia.” In Amran Halim, Lois Carrington, and S. A. Wurm, eds., Papers from the Third International Conference on Austronesian Linguistics, vol. 4: Thematic Variation. Pacific Linguistics, Series C, no. 77. Australian National University, Canberra, pp. 21-33. Kang, Myung-Yoon. 1988. “Topics in Korean Syntax: Phrase Structure, Variable Binding and Movement.” Ph.D. diss., MIT. Kayne, Richard S. 1975. French Syntax. MIT Press, Cambridge, Mass. ———, 1984. Connectedness and Binary Branching. Foris, Dordrecht. —_——. 1989a. “Facets of Romance Past Participle Agreement.” In Paola Beninca’, ed., Dialect

Variation and the Theory of Grammar. Foris, Dordrecht, pp. 85-103. ———. 1989b. “Notes on English Agreement.” CJEFL Bulletin (Hyderabad) 1:41-67. ———. 1989c. “Null Subjects and Clitic Climbing.” In Osvaldo Jaeggli and Ken Safir, eds., The Null Subject Parameter. Kluwer, Dordrecht, pp. 239-261. ———., 1993a. “The Antisymmetry of Syntax.” Unpublished ms., CUNY, Graduate Center. ——_——. 1993b. “Toward a Modular Theory of Auxiliary Selection.” Studia Linguistica 47:3~31. ——_—. 1994. The Antisymmetry of Syntax. MIT Press, Cambridge, Mass. ————. 1997. “The English Complementizer of?’ Journal of Comparative Germanic Linguistics 1:1-11. ——_—. 1998. “Overt vs. Covert Movement’, ms., New York University. Keesing, Roger M. 1985. Kwaio Grammar. Pacific Linguistics, Series B, no. 88. Australian National University, Canberra. Kendall, Martha B. 1976. Selected Problems in Yavapai Syntax (the Verde Valley Dialect). Garland, New York. Kenesei, Istvan, ed. 1990. Approaches to Hungarian, vol. 3. Jate, Szeged. Keyser, Samuel J., ed. 1978. Recent Transformational Studies in European Languages. MIT Press, Cambridge, Mass. Keyser, Samuel J., and Thomas Roeper. 1984. “On the Middle and Ergative Constructions in English.” Linguistic Inquiry 15:355-416. Kinyalolo, Kasangati K. W. 1992. “A Note on Word Order in the Progressive and Prospective in Fon.” Journal of West African Languages 22:35-51. Kiss, Katalin E. 1996. “Two Subject Positions in English.” Linguistic Review 13:119-142. Klima, Edward, and Ursula Bellugi. 1979. The Signs of Language. Harvard University Press, Cambridge, Mass. Klokeid, Terry J. 1969. Thargari Phonology and Morphology. Pacific Linguistics, Series B, no. 12. Australian National University, Canberra. Koizumi, Masatoshi. 1995. “Phrase Structure in Minimalist Syntax.” Ph.D. diss., MIT. Kokora, Dago P. 1976. “Studies in the Grammar of Koyo.” Ph.D. diss., Indiana University (distributed by UMI Dissertation Services, Ann Arbor, Michigan). Koopman, Hilda. 1993. “The Structure of Dutch PPs.” Unpublished ms., University of California at Los Angeles.

Koopman, Hilda, and Claire Lefebvre. 1982. “Pu: marquer de mode, préposition et complémenteur.” In Claire Lefebvre, Héléne Magloire-Holly, and Nanie Pion, eds., Syntaxe de l’Haitien. Ann Arbor, Mich., Karoma, pp. 64-91. Kornfilt, Jaklin. 1985. “Case Marking, Agreement, and Empty Categories in Turkish.” Ph.D. diss., Harvard University. —_——~-, 1996. “On Copular Clitic Forms in Turkish’. Unpublished ms., Syracuse University.

——§., 1997. Turkish. Routledge, London. K6s-Dienes, Dora. 1984. “Sketch of the TMA System of Tigrinya.” In Osten Dahl and Dora Kés-Dienes, eds., Selected Working Papers from the Tense-Mood-Aspect Project, Institute of Linguistics, University of Stockholm, pp. 106-113.

REFERENCES 245

Koster, Jan. 1978. Locality Principles in Syntax. Foris, Dordrecht. Kouwenberg, Silvia. 1994. A Grammar of Berbice Dutch Creole. Mouton de Gruyter, Berlin. Kouwenberg, Silvia, and Ian E. Robertson. 1988. “The Marking of Tense, Mood and Aspect in the Berbice Dutch Creole Language.” In Norbert Boretzky, Werner Enninger, and Thomas Stolz, hrsgg., Beitrdge zum 4. Essener Kolloquium liber “Sprachkontakt, Sprachwandel, Sprachwechsel, Sprachtod.” Studienverlag Dr. N. Brockmeyer, Bochum, pp. 151-174. Kratzer, Angelika. 1977. “What ‘Must’ and ‘Can’ Must and Can Mean.” Linguistics and Philosophy 1:337-355. ——— —, 1981. “The Notional Category of Modality.” In Hans-Jiirgen Eikmeyer and Hannes Rieser, eds., Words, Worlds and Contexts. Mouton de Gruyter, Berlin, pp. 38—74. —————. 1989. “Stage-Level and Individual-Level Predicates.” In Papers on Quantification, report for NSF grant BNS 8719999, Emmon Bach, Angelika Kratzer, and Barbara Partee (principal investigators), Department of Linguistics, University of Massachusetts, Amherst.

Kuno, Susumu. 1975. “Conditions on Verb Phrase Deletion.” Foundations of Language 13:161-175. Ladusaw, William. 1978. “The Scope of Some Sentence Adverbs and Surface Structure.” Proceedings of the 8th Annual Meeting of the North Eastern Linguistic Society, Department of Linguistics, University of Massachusetts, Amherst, pp. 97-111. ———. 1988. “Adverbs, Negation and QR.” In Linguistics in the Morning Calm, 2. Edited by the Linguistic Society of Korea. Hanshin Publishing Company, Seoul, pp. 481-488.

Laenzingler, Christopher. 1993. “Principles for a Formal and Computational Account of Adverbial Syntax.” M.A. thesis, University of Geneva. ——__—. 1996a. “Adverb Syntax and Phrase Structure.” In Anna-Maria Di Sciullo, ed., Configurations: Essays on Structure and Interpretation. Cascadilla Press, Somerville, Mass.,

pp. 99-127. —_——. 1996b. “Comparative Studies in Word Order Variations: Adverbs, Pronouns and Clause Structure in Romance and Germanic.” Ph.D. diss., University of Geneva. Laka, Itziar. 1990. “Negation in Syntax: On the Nature of Functional Categories and Projections.” Ph.D. diss., MIT. ———. 1993. “The Structure of Inflection: A Case Study in X° Syntax.” In Jose Ignacio Hualde and Jon Ortiz de Urbina, eds., Generative Studies in Basque Linguistics. John Benjamins, Amsterdam, pp. 21—70. Lang, Ewald. 1979. “Zum Status der Satzadverbiale.” Slovo a Slovesnost 40 (1—3):200—-213. Landerman, Peter, and Donald Frantz. 1972. Notes on Grammatical Theory. Summer Institute of Linguistics (in cooperation with the Ministry of Education), Lima, Peru. Larochette, Joe. 1980. “La notion d’aspect: le point de vue d’un africaniste.” In Jean David and Robert Martin, eds., La notion d’aspect. Klincksieck, Paris, pp. 31-40. Larson, Richard. 1985. “Bare NP-Adverbs.” Linguistic Inquiry 16:595-621.

——. 1988. “On the double-object construction”, Linguistic Inquiry, 19.335-391. ———. 1990. “Double Objects Revisited: Reply to Jackendoff.” Linguistic Inquiry 21:589623. Lawrence, Marshall. 1972. “Oksapmin Sentence Structure.” Papers in New Guinea Linguistics 16:17—-46. Pacific Linguistics, Series A, no. 34. Australian National University, Canberra. Leblanc, Annie. 1989. “Le statut syntaxique des adverbes de temps en créole haitien.” Revue Québécoise de Linguistique 18:41-63. Lefebvre, Claire. 1995. “Les marqueurs préverbaux du fongbe et du créole haitien: étude de sémantique comparative.” Linguistique Africaine 14:155—180. ———. 1996. “The Tense, Mood and Aspect System of Haitian Creole and the Problem of

Transmission of Grammar in Creole Languages.” Journal of Pidgin and Creole Languages 11.231-311.

246 REFERENCES

Lefebvre, Claire, Héléne Magloire-Holly, and Nanie Piou, eds. 1982. Syntaxe de |’ Haitien. Ann Arbor, Mich., Karoma. Lees, Robert B. 1960. The Grammar of English Nominalizations. Mouton, The Hague. Leitch, Miles. 1994. “Babole.” In Peter Kahrel and René van den Berg, eds., Typological Studies in Negation. John Benjamins, Amsterdam, pp. 190-210. Lepschy, Anna Laura, and Giulio Lepschy. 1977. The Italian Language Today. Hutchinson, London. Lewis, David. 1975. “Adverbs of Quantification.” In Edward Keenan, ed., Formal Semantics of Natural Language. Cambridge University Press, Cambridge, pp. 3—15. Lewis, Sandra C. 1972. “Sanio-Hiowe Verb Phrases.” Papers in New Guinea Linguistics 15:11-22. Pacific Linguistics, Series A, no. 31. Australian National University, Canberra. Li, Charles N. 1991. “The Aspectual System of Hmong.” Studies in Language 15:25-58. Lightfoot, David, and Norbert Hornstein, eds. 1994, Verb Movement. Cambridge University Press, Cambridge.

Lindstedt, Jouko. 1985. On the Semantics of Tense and Aspect in Bulgarian. Slavica Helsingiensia, no. 4. University of Helsinki, Helsinki. Lois, Ximena. 1989. “Aspects de la syntaxe de l’ espagnol et théorie de la grammaire.” Ph.D. diss. (nouveau régime), University of Paris VIII. Lonzi, Lidia. 1990. “Which Adverbs in Spec, VP?” Rivista di Grammatica Generativa 15:141-160. ———. 1991. “Ii sintagma avverbiale.” In Lorenzo Renzi and Giampaolo Salvi, eds., Grande grammatica italiana di consultazione. Vol. 2. 1 Mulino, Bologna, pp. 342-412. Louw, J. A. 1987. “Auxiliary Verb in Xhosa.” South African Journal of African Languages 7(1):7-15. Louwerse, J. 1988. The Morphosyntax of Una in Relation to Discourse Structure: A Descriptive Analysis. Pacific Linguistics, Series B, no. 100. Australian National University, Canberra. Louwrens, Louis J., Ingeborg M. Kosch, and Albert E. Kotzé. 1995. Northern Sotho. Lincom Europa, Munich. Lynch, John. 1982. “Anejom Grammar Sketch.” In John Lynch, ed., Papers in Linguistics of Melanesia. 4:93-154. Pacific Linguistics, Series A, no. 64. Australian National University, Canberra. Lyons, John. 1968. Introduction to Theoretical Linguistics. Cambridge University Press, Cambridge. ———. 1977. Semantics. Cambridge University Press, Cambridge. MacDonald, Lorna. 1990. A Grammar of Tauya. Mouton de Gruyter, Berlin. Magloire-Holly, Héléne. 1982. “Les modaux: auxiliaires ou verbes?” In Claire Lefebvre, Héléne Magloire-Holly, and Nanie Piou, eds., Syntaxe de |’Haitien. Ann Arbor, Mich., Karoma, pp. 92-121. Mahajan, Anoop. 1989. “Agreement and Agreement Phrases.” MIT Working Papers in Linguistics 10:217-252. ——_—. 1990. “The A/A-Bar Distinction and Movement Theory.” Ph.D. diss., MIT. Mahootian, Shahrzad, and Beatrice Santorini. 1996. “Code Switching and the Complement/ Adjunct Distinction.” Linguistic Inquiry 27:464-479., Matthews, G. H. 1965. Hidatsa Syntax. Mouton, The Hague.

Matthews, Stephen, and Virginia Yip. 1994. Cantonese: A Comprehensive Grammar. Routledge, London. May, Robert. 1985. Logical Form: Its Structure and Derivation. MIT Press, Cambridge, Mass. McCawley, James D. 1988. ‘“Adverbial NPs: Bare or Clad in See-through Garb.” Language

64:583-590.

REFERENCES 247

——_——, 1990. “Remarks on Adverbial Constituent Structure.” In Carol Georgopoulos and Roberta Ishihara, eds., Interdisciplinary Approaches to Language: Essays in Honor of S.-Y¥.Kuroda. Kluwer, Dordrecht, pp. 415-433. McCloskey, James. 1992. “Adjunction, Selection and Embedded Verb Second.” Ms., University of California at Santa Cruz. ——., 1996. “On the Scope of Verb Movement in Irish.” Natural Language and Linguistic Theory 14:47-104. ———. 1997. “Subjecthood and Subject Positions.” In Liliane Haegeman, ed., Elements of Grammar, Kluwer, Dordrecht, pp. 197-235. McConnell-Ginet, Sally. 1982. “Adverbs and Logical Form.” Language 58:144—184. Mchombo, Sam. 1986. “The Non-existence of Verb Object Agreement in Bantu.” Unpublished ms., San Jose State University, San Jose, Calif. McLendon, Sally. 1996. “Sketch of Eastern Pomo, a Pomoan Language’, in Ives Goddard, ed., 1996. Languages (vol. 17 of Handbook of North American Indians). Smithsonian Institution, Washington, pp. 307-550.

Mennecier, Philippe, and Nicole Tersis. 1988. “Remarques préliminaires sur le temps et )’ aspect en tunumiisut.” In Nicole Tersis and Alain Kihm, eds., Temps et aspect. Actes du colloque C.N.R.S., Paris, October 24—25, 1985. Peeters/ SELAF, Paris, pp. 197204. Michaelis, Laura A. 1991. “Temporal Priority and Pragmatic Ambiguity: The Case of Already.” Proceedings of the 17th Annual Meeting of the Berkeley Linguistics Society. Linguistics Society, Berkeley, Calif., pp. 426-438. —_——., 1993. “’Continuity’ within Three Scalar Models: The Polysemy of Adverbial Still.” Journal of Semantics 10:193—237. Michaelis, Susanne. 1993. Temps et aspect en créole seychellois: valeurs et interférences. Helmut Buske Verlag, Hamburg. Michell, Gillian. 1975. “A Pragmatic Analysis of Sentence Adverbs: A Study with Particular Reference to Obviously.” UMI Dissertation Services, Ann Arbor, Mich. Mitchell, Erika. 1991. “Evidence from Finnish for Pollock’s Theory of IP.” Linguistic Inquiry 22:373-379. —___—. 1993. “Agreement and Negation in the Baltic Finnish Languages.” In Loren Smith

Stvan, ed., Papers from the 3’ Annual Meeting of the Formal Linguistics Society of Midamerica. Distributed by LU.L.C., Bloomington, Ind., pp. 235~252. ———., 1994, “On the Position of NegP in English and the Status of Not.” Cornell Working Papers in Linguistics 12:94-126. Mithun, Marianne. Forthcoming. The Languages of Native North America. Cambridge University Press, Cambridge. Mithun, Marianne, and Elisabeth Ali. 1996. “The Elaboration of Aspectual Categories: Central Alaskan Yup’ik.” Folia Linguistica 30:111—127. Mixco, Mauricio. 1997. Mandan. Lincom Europa, Munich. Moltmann, Friederike. 1989. “Measure Adverbials as Part Quantifiers.” Proceedings of the 8th West Coast Conference on Formal Linguistics. Stanford Linguistics Association, Stanford, Calif., pp. 247-261. —_—-, 1990. “Scrambling in German and the Specificity Effect.” Unpublished ms., MIT.

——. 1991. “Measure Adverbials.” Linguistics and Philosophy 14:629-660. —_——. 1995. “Part-Structure Modification.” Unpublished ms., CUNY, Graduate Center. Montague, Richard M. 1974. Formal Philosophy. Ed. R. H. Thomason. Yale University Press, New Haven, Conn. Moravesik, Edith, and Jessica Wirth. 1986. “Markedness—An Overview.” In Fred Eckman, Edith Moravesik, and Jessica Wirth, eds., Markedness. Plenum Press, N.Y., pp. 1-11.

248 REFERENCES

Merdrup, Ole. 1976. “Une analyse non-transformationnelle des adverbes en -ment.” Revue Romane 11:317-333. Mosel, Ulrike, and Even Hovdhaugen. 1992. Samoan Reference Grammar. Scandinavian University Press, Oslo. Motapanyane, Virginia. 1995. Theoretical Implications of Complementation in Romanian. Unipress, Padua. Mourelatos, Alexander P. D. 1981. “Events, Processes and States.” In Philip Tedeschi and Annie Zaenen, eds., Tense and Aspect. Vol. 14 of Syntax and Semantics. Academic Press, New York, pp. 191-212. Mugane, John M. 1997. A Paradigmatic Grammar of Giktiyii. CSLI Publications, Stanford. Muller, Claude. 1991. La négation en francais. Librairie Droz, Geneva. Miller, Gereon, and Wolfgang Sternefeld. 1993. “Improper Movement and Unambiguous Binding.” Linguistic Inquiry 24:461—507. Muysken, Pieter. 1977. Syntactic Developments in the Verb Phrase of Ecuadorian Quechua. Peter de Ridder Press, Lisse. —__——, ed. 1981. Generative Studies on Creole Language. Foris, Dordrecht. Myhill, John. 1992. Typological Discourse Analysis. Blackwell, Oxford. Napoli, Donna Jo. 1974. “The No Crossing Filter.” Papers from the 10th Regional Meeting of the Chicago Linguistic Society. Chicago Linguistic Society, Chicago, pp. 482-491. Nedyalkov, Igor. 1994. “Evenki.” In Peter Kahrel and René van den Berg, eds., Typological Studies in Negation. John Benjamins, Amsterdam, pp. 1—34. ———. 1997. Evenki. Routledge, London. Newman, Paul. 1992. “Chadic Languages.” In William Bright, ed., International Encyclopedia of Linguistics. Oxford University Press, New York, pp. 253-254. Neguendjio, Emile-Gille. 1992. “Temps et aspect de la langue Bangwa.” Journal of West African Languages 22(2):89-100. Nichols, Lynn. 1993. “Recovering Zuni Auxiliaries and Their Role in Event Classification.” Harvard Working Papers in Linguistics 2:92—108. Nida, Eugene. 1974. Morphology. University of Michigan Press, Ann Arbor. Nilsen, @ystein. 1997. “Adverbs and A-shift.” Working Papers in Scandinavian Syntax 59:1—31. Department of Scandinavian Languages, University of Lund. —_——. 1998. The Syntax of Circumstantial Adverbials. ms., University of Tromsg. Nglke, Henning. 1990. “Recherches sur les adverbes: bref apergu historique des travaux de classification.” Langue Francaise 88:1 17-127. Nuyts, Jan. 1993. “Epistemic Modal Adverbs and Adjectives and the Layered Representation of Conceptual and Linguistic Structure.” Linguistics 31:933-969. Nuyts, Jan, A., Machtelt Bolkenstein, and Co Vet, eds. 1990. Layers and Levels of Representation in Language Theory. John Benjamins, Amsterdam. O Baoill, Dénall P. 1994. “Tense and Aspect in Modern Irish.” In Rolf Thieroff and Joachim Ballweg, eds., Tense Systems in European Languages. Niemeyer, Tiibingen, pp. 201216. Obenauer, Hans G. 1994. “Aspects de la syntaxe A-barre.” Ph.D. diss., University of Paris VIII. O’Connor, Mary C. 1992. Topics in Northern Pomo Grammar. Garland, New York. Ofuani, Ogo A. 1981. “Future Time Expression in Nigerian Pidgin.” Papers in Linguistics 14(3):309-325. ———. 1982. “A Discussion of the Progressive Aspect in Nigerian Pidgin.” Papers in Linguistics 15(3):229-239. Ojea Lopez. 1994. “Adverbios y categorias funcionales en espafiol.” Revista Espanola de Lingitistica 24:393-416.

REFERENCES 249

Oke, David O. 1972. “On the Construction and Semantic Interpretation of Auxiliary Clusters in Yoruba.” In A. Bamgbose, ed., The Yoruba Verb Phrase. Ibadan University Press, Ibadan, pp. 135-161. Omamor, A. P. 1982. “Tense and Aspect in Isekiri.” Journal of West African Languages 12(2):95-129. Omoruyi, Thomas O. 1991. “Tense, Aspect and Modality in Edo”, in Afrika und Ubersee

74:1-19. Ord6jfiez, Francisco. 1994, “Postverbal Asymmetries in Spanish.” Glow Newsletter 32:40—-41.

Ortiz de Urbina, Jon. 1989. Parameters in the Grammar of Basque. Foris, Dordrecht. Osborn, Henry A. Jr. 1967. “Warao III: Verbs and Suffixes.” International Journal of American Linguistics 33:46—64. Quhalla, Jamal. 1988. “The Syntax of Head Movement: A Study of Berber.” Ph.D. diss., University College, London. ———, 1990. “Sentential Negation, Relativized Minimality and the Aspectual Status of Auxiliaries.” Linguistic Review 7:183-231. ———. 1991. Functional Categories and Parametric Variation. Routledge, London. ——_——. 1993. “Negation, Focus and Tense: The Arabic maa and laa.” Rivista di Linguistica 5:275-300. Oyharcabal, Bernard. 1993. “Verb Agreement with Non Arguments: On Allocutive Agreement.” In Jose Ignacio Hualde and Jon Ortiz de Urbina, eds., Generative Studies in Basque Linguistics. John Benjamins, Amsterdam, pp. 89-114. Palmer, Frank R. 1986. Mood and Modality. Cambridge University Press, Cambridge.

Paoli, Sandra. 1997. “Agreement: A Relational Property or a Functional Projection? Evidence from Past Participle Agreement in Friulian.” M.A. thesis, University of York. Papen, Robert Antoine. 1978. “The French-based Creoles of the Indian Ocean: An Analysis and a Comparison.” Ph.D. diss., University of California at San Diego (distributed by UMI Dissertation Services, Ann Arbor, Mich.). Parry, Mair. 1997. “Negation.” In Martin Maiden and Mair Parry, eds., The Dialects of Italy. Routledge, London, pp. 179-185. Parsons, Terence. 1990. Events in the Semantics of English: A Study in Subatomic Semantics. MIT Press, Cambridge, Mass. Payne, David. 1981. The Phonology and Morphology of Axininca Campa. Summer Institute of Linguistics and University of Texas at Arlington. Payne, John R. 1985. “Negation.” In Timothy Shopen, ed., Language Typology and Syntactic Description. Cambridge University Press, Cambridge, vol. 1, pp. 197-242. Pecoraro, Walter, and Chiara Pisacane. 1984. L’Avverbio. Zanichelli, Bologna. Peeke, Catherine. 1994. “Waorani.” In Peter Kahrel and René van den Berg, eds., Typological Studies in Negation. John Benjamins, Amsterdam, pp. 267—290. Perkins, Michael R. 1983. Modal Expressions in English. Ablex, Norwood, N.J. Pesetsky, David. 1989. “Language-particular Processes and the Earliness Principle”, Unpublished ms., MIT, Cambridge, Mass.

Phillips, D. J. 1976. Wahgi Phonology and Morphology. Pacific Linguistics, Series B, no. 36. Australian National University, Canberra. Picallo, Carme M. 1985. Opaque Domains, Ph.D. Dissertation, City University of New York. ———.. 1990. “Modal Verbs in Catalan.” Natural Language and Linguistic Theory 8:285312. Pittner, Karin. 1995. “Adverbiale im Deutschen.” Unpublished ms., University of Stuttgart. Platzack, Christer. 1979. The Semantic Interpretation of Aspect and Aktionsarten: A Study of Internal Time Reference in Swedish, Foris, Dordrecht.

250 REFERENCES

——_——. 1986. “The Position of the Finite Verb in Swedish.” In Hubert Haider and Martin

Prinzhorn, eds., Verb Second Phenomena in Germanic Languages. Foris, Dordrecht, pp. 27-47. ———. 1994. “Mainland Scandinavian Evidence for Split COMP.” Unpublished ms., University of Lund. Poletto, Cecilia. 1993. “The Aspect Projection: An Analysis of the ‘Passé Surcomposé’.” In Elisabetta Fava, ed., Proceedings of the 17th Meeting of Generative Grammar. Rosenberg and Sellier, Turin, pp. 289-311.

———. 1997, “Rhaetoromance Verb Second: A Split CP Perspective.” Unpublished ms., Centro di Dialettologia del CNR, Padua. ——_——. 1998. “Two Types of Emphatic Adverbs in Northern Italian Varieties.” Unpublished

ms., Centro di Dialettologia del CNR, Padua. Pollock, Jean-Yves. 1989. “Verb Movement, Universal Grammar and the Structure of IP.” Linguistic Inquiry 20:365—424. —_——., 1997. “Notes on Clause Structure.” In Liliane Haegeman, ed. Elements of Grammar, Kluwer, Dordrecht, pp. 237-279. Popjes, Jack, and Jo Popjes. 1986. “Canela-Crahé.” In Desmond C. Derbyshire and Geoffrey K. Pullum, eds., Handbook of Amazonian Languages. Mouton de Gruyter, Berlin, vol.1, pp. 128-199. Portner, Paul, and Raffaella Zanuttini. 1996. “The Force of Negation in Wh Exclamatives and Interrogatives.” Unpublished ms., Georgetown University. Potet, Jean-Paul G. 1992. “An Adverbial-to- Verbal Morpheme Transfer in Classical Tagalog.” Lingua 86:1-46. Pozzobon, Samantha. 1998. “Zur Distribution der Adverbiale im Deutschen und Japanischen.” Tesi di laurea, University of Venice. Puskas, Genoveva. 1996. “Word Order in Hungarian: The Syntax of A-Bar Positions.” Ph.D. diss., University of Geneva. Quirk, Randolf, Sidney Greenbaum, Geoffrey Leech, and Jan Svartvik. 1985.A Comprehensive Grammar of the English Language. Longman, London. Rackowski, Andrea. 1996. “Malagasy Adverbs”. M.A. diss., McGill University. Ramat, Paolo. 1994. “Osservazioni sugli avverbi di frase.” In P. Cipriano et al., eds., Miscellanea di studi in onore di Walter Belardi. 11 Calamo, Rome, pp. 897-915. Ramat, Paolo, and Davide Ricca. 1994, “Prototypical Adverbs: On the Scalarity/Radiality of the Notion apverB.” Rivista di Linguistica 6:289-326. Ramchand, Gillian Catriona. 1995. “Aspect and Predication: The Semantics of Argument Structure.” Unpublished ms., Oxford University. Refsing, Kirsten. 1994. “Tense, Aspect and Actionality in the Ainu Language.” In Carl Bache, Hans Basbgll, and Carl-Erik Lindberg, eds., Tense Aspect and Action: Empirical and Theoretical Contributions to Language Typology. Mouton de Gruyter, Berlin, pp. 311-321. Rehg, Kenneth L. 1981. Ponapean Reference Grammar. University of Hawaii Press, Honolulu. Reichenbach, Hans. 1947. Elements of Symbolic Logic. Free Press, New York. Reinhart, Tanya. 1995. “Interface Strategies.” OTS, Utrecht University. Renzi, Lorenzo, ed. 1988. Grande grammatica italiana di consultazione, vol. 1. Il Mulino, Bologna.

Renzi, Lorenzo, and Giampaolo Salvi, eds. 1991. Grande grammatica italiana di consultazione, vol. 2. Il Mulino, Bologna. Rescher, Nicolas. 1968. Topics in Philosophical Logic. Reidel, Dordrecht. Reuland, Eric. 1988. “Indefinite Subjects.” Proceedings of the 18th Annual Meeting of the North Eastern Linguistic Society, Department of Linguistics, University of Massachusetts, Amherst, pp. 375—394.

REFERENCES 251

Rice, Keren. 1989. A Grammar of Slave. Mouton de Gruyter, Berlin. ———. 1991a. “On Deriving Rule Domains: The Athapaskan Case.” Proceedings of the 10th West Coast Conference on Formal Linguistics. Stanford Linguistic Association, Stanford, Calif., pp. 417-430. ——. 1991b. “Predicting the Order of the Disjunct Morphemes in the Athapaskan Languages.” Toronto Working Papers in Linguistics 11(1):99~121. Riemsdijk, Henk van. 1978. A Case Study in Syntactic Markedness. Peter de Ridder Press, Lisse. ———. 1990. “Functional Prepositions.” In Harm Pinkster and I. Genée, eds., Unity in Diversity. Foris, Dordrecht, pp. 229-241. —_——. 1996. “Categorial Feature Magnetism: The Extension and Distribution of Projections.” Unpublished ms., Tilburg University. Rivero, Maria-Luisa. 1990. “The Location of Nonactive Voice in Albanian and Modern Greek.” Linguistic Inquiry 21:134-146. ———. 1992. “Adverb Incorporation and the Syntax of Adverbs in Modern Greek.” Linguistics and Philosophy 15:289-331. ———, 1994. “Clause Structure and V-Movement in the Languages of the Balkans.” Natural Language and Linguistic Theory 12:63-129. Rizzi, Luigi. 1978. “A Restructuring Rule in Italian Syntax.” In Samuel J. Keyser, ed., Recent Transformational Studies in European Languages. MIT Press, Cambridge, Mass., pp. 113-158. ——_———. 1981. “Nominative Marking in Italian Infinitives.” In Frank Heny, ed., Binding and

Filtering. Croom Helm, London, pp. 129-157. ———., 1982. Issues in Italian Syntax. Foris, Dordrecht. ———-. 1990. Relativized Minimality. MIT Press, Cambridge, Mass. ———. 1991. “Residual Verb Second and the Wh-Criterion.” Technical Reports in Formal and Computational Linguistics, no. 2, University of Geneva. ———, 1997. “The Fine Structure of the Left Periphery,” in Liliane Haegeman, ed., Elements of Grammar, Kluwer, Dordrecht, pp. 281-337. Roberts, Ian. 1985a. “Agreement Parameters and the Development of English Modal Auxiliaries.” Natural Language and Linguistic Theory 3:21-58. ———. 1985b. “The Representation of Implicit and Dethematized Subjects.” Ph.D. diss., University of Southern California. ———. 1987. The Representation of Implicit and Dethematized Subjects. Foris, Dordrecht. ———. 1993. Verbs and Diachronic Syntax. A Comparative History of English and French. Kluwer, Dordrecht. —_——. 1995. “Object Movement and Verb Movement in Early Modern English.” In Hubert Haider, Susan Olsen, and Sten Vikner, eds., Studies in Comparative Germanic Syntax. Kluwer, Dordrecht, pp. 269-284. Robertson, Ian E. 1990. “The Tense-Mood-Aspect System of Berbice Dutch.” In John Victor

Singler, ed., Pidgin and Creole Tense-Mood-Aspect Systems. John Benjamins, Amsterdam, pp. 169-184. Robinson, Lila W., and James Armagost. 1990. Comanche Dictionary and Grammar. Summer Institute of Linguistics and the University of Texas at Arlington. Rochette, Anne. 1990. “La structure d’arguments et les propriétés distributionnelies des adverbes.” Revue Québécoise de Linguistique 20:55-77. Rohrer, Christian, ed. 1977. On the Logical Analysis of Tense and Aspect. Gunter Narr, Tiibingen. Rooth, Mats. 1985. “Association with Focus.” Ph.D. diss., University of Massachusetts at Amherst,

252 REFERENCES

Rose, Sharon. 1996. “Inflectional Affix Order in Ethio-Semitic.” In Jacqueline Lecarme, Jean Lowenstamm, and Ur Shlonsky, eds., Studies in Afroasiatic Grammar, Holland Academic Graphics, The Hague, pp. 337—353. Rottet, Kevin J. 1992. “Functional Categories and Verb Movement in Louisiana Creole.” Probus 4:261-289. Ruhlen, Merritt. 1991. Classification. Vol. 1 of A Guide to the World’s Languages. Edward Amold, London. Ruwet, Nicolas. 1968. Introduction a la grammaire générative. Plon, Parts. Sadock, Jerrold N., and Arnold M. Zwicky. 1985. “Speech Act Distinctions in Syntax.” In Timothy Shopen, ed., Language Typology and Syntactic Description. Cambridge University Press, Cambridge, vol. 1, pp. 155-196. Sapir, Edward. 1921. Language. Harcourt Brace, New York. Saxena, Anju. 1995. “New Aspect Morphology: Where Does It Come From?” Reports from Uppsala University Linguistics 28:1-20. Schaaik, Gerjan van. 1994. “Turkish.” In Peter Kahrel and René van den Berg, eds., Typological Studies in Negation. John Benjamins, Amsterdam, pp. 35—50. Schachter, Paul. 1985. “Parts-of-Speech Systems.” In Timothy Shopen, ed., Language Ty-

pology and Syntactic Description. Cambridge University Press, Cambridge, vol. 1, pp. 3-61. Schlyter, Suzanne. 1974. “Une hiérarchie d’adverbes et leur distributions—par quelles transformations?” In Christian Rohrer and Nicolas Ruwet, eds., Actes du colloque francoallemand de grammaire transformationnelle, vol. 2, Niemeyer, Tiibingen, pp. 76-84.

———, 1977. “La place des adverbes en -ment en francais.” Ph.D. diss., University of Konstanz. Schreiber, Peter A. 1971. “Some Constraints on the Formation of English Sentence Adverbs.” Linguistic Inquiry 2:83-101. ——_—.. 1972. “Style Disjuncts and the Performative Analysis.” Linguistic Inquiry 3:321-—348.

Schiitze, Carson T. 1995. “Case Verb Morphology and Argument Structure in Choctaw: A Minimalist Account.” Proceedings of the 25th North Eastern Linguistic Society, Department of Linguistics, University of Massachusetts, Amherst, pp. 441-455. Seuren, Peter A. 1981 “Tense and Aspect in Sranan.” Linguistics 19:1043—1076. ———., 1983 “The Auxiliary System in Sranan.” In Frank Heny and Barry Richards, eds., Linguistic Categories: Auxiliaries and Related Puzzles, vol. 2. Reidel, Dordrecht, pp. 219-251. ————. 1995 “Notes on the History and Syntax of Mauritian Creole.” Linguistics 33:531-577.

Shlonsky, Ur. 1991. “Quantifiers as Functional Heads: A Study of Quantifier Float in Hebrew.” Lingua 84:159-180. ————-, 1997. Clause Structure and Word Order in Hebrew and Arabic. Oxford University Press, New York. Shopen, Timothy, ed. 1985. Language Typology and Syntactic Description. Cambridge University Press, Cambridge. Siewierska, Anna. 1991. Functional Grammar. Routledge, London. ———_-—. 1992. “Layers in FG and GB.” In Michael Fortescue, Peter Harder, and Lars Kristoffersen, eds., Layered Structure and Reference in a Functional Perspective. John Benjamins, Amsterdam, pp. 409-4372. Sigurdsson, Halldér Armann. 1989. “Verbal Syntax and Case in Icelandic: In a Comparative GB Approach.” Ph.D. diss., University of Lund. Singh, Mona. 1991. “The Perfective Paradox: Or How to Eat Your Cake and Have It Too.” Proceedings of the 17th Annual Meeting of the Berkeley Linguistics Society. Linguistics Society, Berkeley, Calif., pp. 469-479.

REFERENCES 253

Singler, John Victor, ed., 1990. Pidgin and Creole Tense-Mood-Aspect Systems. John Benjamins, Amsterdam. Smith, Carlota. 1987. “Notes on Aspect in Chinese.” Texas Linguistic Forum 28:91-127. ———., 1991. The Parameter of Aspect. Kluwer, Dordrecht. Sneddon, James Neil. 1996. Indonesian Reference Grammar. Allen & Unwin, St. Leonards, Australia. Sohn, Ho-Min. 1994. Korean. Routledge, London. Sohn, Ho-Min, and B. W. Bender. 1973. A Ulithian Grammar. Pacific Linguistics, Series C, no. 27. Australian National University, Canberra. Song, Jae-Mog. 1995. “Progressives in Korean and Khalkha Mongolian.” Working Papers in Linguistics and Phonetics. School of Oriental and African Studies, University of London, 5:251-270. Spaulding, Craig and Pat Spaulding. 1994. Phonology and Grammar of Nankina (Data Papers on Papua New Guinea Languages, vol. 41). Summer Institute of Linguistics, Ukarumpa via Lae, Papua, New Guinea. Spears, Arthur K. 1989. “AP and VA Futures in Haitian Creole.” CUNYForum 14:204-210. ——., 1990. “Tense, Mood and Aspect in the Haitian Creole Preverbal Marker System.” In John Victor Singler, ed., Pidgin and Creole Tense-Mood-Aspect Systems. John Benjamins, Amsterdam, pp. 119-142. Speas, Margaret. 1991a. “Functional Heads and Inflectional Morphemes.” Linguistic Review 8:389-417. ———. 1991b. “Functional Heads and the Mirror Principle.” Lingua 84:181-214. Spellmire, Susan. 1994, “The Position of Modals.” Unpublished ms., University of California at Los Angeles. Sportiche, Dominique. 1988. “A Theory of Floating Quantifiers and Its Corollaries for Constituent Structure.” Linguistic Inquiry 19:425-449. ————. 1992. “Clitic Constructions.” Unpublished ms., University of California at Los Angeles. —_——, 1993. “Adjuncts and Adjunction.” Unpublished ms., University of California at Los Angeles. ————. 1995a. “French Predicate Clitics and Clause Structure.” In Anna Cardinaletti and Maria Teresa Guasti, eds., Small Clauses, Academic Press, New York, pp. 287-324. ——_——. 1995b. “Sketch of a Reductionist Approach to Syntactic Variation and Dependencies.” In Héctor Campos and Paula Kempchinsky, eds., Evolution and Revolution in Linguistic Theory. Georgetown University Press, Washington, D.C., pp. 356-398. ——. 1996. “Clitic Constructions.” In Johan Rooryck and Laurie Zaring, eds., Phrase Structure and the Lexicon. Kluwer, Dordrecht, pp. 213-276. Starke, Michal. 1995. “On the Format for Small Clauses.” In Anna Cardinaletti and Maria Teresa Guasti, eds., Small Clauses. Academic Press, New York, pp. 237-269.

Stechow, Arnim von. 1996. “The Different Readings of Wieder ‘Again’: A Structural Account.” Journal of Semantics 13:87-138. Steele, Susan. 1975. “Is It Possible?” Working Papers on Language Universals. Stanford University, Stanford, Calif., 18:35~58. Steffensen, Margaret S. 1989. “Gen: An Affected/Evidential Particle in Australian Creole English.” Papers from the 25" Regional Meeting of the Chicago Linguistic Society, Chicago Linguistic Society, Chicago, pp. 254-265. Steinitz, Renate. 1969. Adverbial-Syntax. Studia Grammatica 10. Akademie Verlag, Berlin. Sterner, Robert. 1975. “Sobei Verb Inflection.” Oceanic Linguistics 14:128-145. Stock, Roberta. 1973. “On Recently and Lately.” Studies in the Linguistic Sciences 3(1):231248.

254 REFERENCES

Stowell, Tim. 1995. “What Do the Present and Past Tenses Mean?” In Piermarco Bertinetto et al., eds., Temporal Reference, Aspect and Actionality, vol. 1. Rosenberg and Sellier, Turin, pp. 381-396.

—. 1996. “The Phrase Structure of Tense.” In Johan Rooryck and Laurie Zaring, eds., Phrase Structure and the Lexicon. Kluwer, Dordrecht, pp. 277-291. Stroik, Thomas. 1990. “Adverbs as V-sisters”, Linguistic Inquiry, 21.654—61. ———. 1992. “On the Distribution of Temporal and Locative NP Adverbials.” Linguistic Review 9:267-284. Sueur, Jean-Pierre. 1978. “Adverbes de modalité et verbes modaux épistémiques.” Recherches Linguistiques 5/6:235—272. Sufier, Margarita. 1987. “Haber + Past Participle.” Linguistic Inquiry 18:683-690. Svantesson, Jan-Olof. 1991. “Tense, Mood and Aspect in Mongolian.” Working Papers. Department of Linguistics, University of Lund, 38:189-—204. —————, 1994. “Tense, Mood and Aspect in Kammu.” In Carl Bache, Hans Basbgll, and Car!-

Erik Lindberg, eds., Tense Aspect and Action: Empirical and Theoretical Contributions to Language Typology. Mouton de Gruyter, Berlin, pp. 265—278. Sylvain, Suzanne. 1979. Le Créole haitien: morphologie et syntaxe. Slatkine Reprints, Geneva. Taraldsen, Tarald. 1983. “Parametric Variation in Phrase Structure: A Case Study.” Ph.D. diss., University of Tromsg. Taylor, Charles. 1985. Nkore-Kiga. Croom Helm, London. Tedeschi, Philip, and Annie Zaenen, eds. 1981. Tense and Aspect. Vol. 14 of Syntax and Semantics. Academic Press, New York. Tenny, Carol. 1987. “Grammaticalizing aspect and affectedness.” Ph.D. diss., MIT, Cambridge, Mass. Tesniére, Lucien. 1939. “Théorie structurale des temps composés.” In M. Sechehaye et al., eds., Mélanges de linguistique offerts a Charles Bally, Georg, Geneva, pp. 153-183. Thomason, Richmond, and Robert Stalnaker. 1973. “A Semantic Analysis of Adverbs.” Linguistic Inquiry 4:195-220. Thompson, Laurence C. 1965. A Vietnamese Grammar. University of Washington Press, Seattle. Thrainsson, Héskuldur, and Sten Vikner. 1995. “Modals and Double Modals in the Scandinavian Languages.” Working Papers in Scandinavian Syntax 55:51-—88. Department of Scandinavian Languages, University of Lund. Thurgood, Eizbieta, and Graham Thurgood. 1996. “Aspect, Tense or Aktionsart? The Particle JA in Kristang (Malacca Creole Portuguese).” Journal of Pidgin and Creole Languages 11:45-70. Thurgood, Graham. 1986. ““The Nature and Origin of the Akha Evidentials System.” In Wallace Chafe and Johanna Nichols, eds., Evidentiality: The Linguistic Coding of Epistemology. Ablex, Norwood, N.J., pp. 214-222. Togeby, Knud. 1984. Les mots invariables. Vol. 4 of Grammaire francaise. Akademisk Forlag, Copenhagen. Tovena, Lucia. 1996. “The Context Sensitivity of the Italian Adverb ancora.” In Antonietta Bisetto et al., eds., Proceedings of ConSOLE, Ill. Holland Academic Graphics, The Hague, pp. 231-246. Tracy, Frances V. 1974. “An Introduction to Wapishana Verb Morphology.” Jnternational Journal of American Linguistics 40:120-125. Traugott, Elizabeth C., and John Waterhouse. 1969. “‘Already’ and ‘Yet’: A Suppletive Set of Aspect Markers.” Journal of Linguistics 5:287-304. Travis, Lisa. 1988. “The Syntax of Adverbs.” In McGill Working Papers in Linguistics. Special Issue on Comparative Germanic Syntax. Department of Linguistics, McGill University, pp. 280-310.

REFERENCES 255

Troike, Rudolph C. 1996. “Sketch of Coahuilteco, a Language Isolate of Texas”, in Ives Goddard (ed.) Languages (vol. 17 of Handbook of North American Indians). Smithsonian

Institution, Washington, pp. 644-665. Tryon, D. T. 1970. An Introduction to Maranungku (Northern Australia). Pacific Linguistic, Series B, no. 15. Australian National University, Canberra. Tsai, Mei-chih. 1995. La modification adverbiale et l’ordre des mots en chinois. Ph.D. diss., University of Paris 7. Tura, Sabahat Sausa. 1981. “‘Yes, He Hasn’t’ and a Few Other Not’s in Turkish.” In Proceedings of the 7th Annual Meeting of the Berkeley Linguistics Society. Linguistics Society, Berkeley, Calif., pp. 317-327.

Uhlenbeck, C. C. 1938. A Concise Blackfoot Grammar. N.V. Noord-Hollandische Uitgeversmaatschappij, Amsterdam. van den Berg, Helma. 1995. A Grammar of Hunzib. Lincom Europa, Munich. van der Auwera, Johan, ed. 1991. “Adverbs and Particles of Change and Continuation.” Eurotyp Working Paper, Series V, no. 2. Department of Linguistics, University of Antwerp. —-_——. 1993. “‘Already’ and ‘Sull’: Beyond Duality.” Linguistics and Philosophy 16:613— 653. ——. Forthcoming. “Phasal Adverbials in the Languages of Europe.” In Johan van der Auwera and Donall P. O Baoill, eds., Adverbial Constructions in the Languages of Europe, Mouton de Gruyter, Berlin. Vangsnes, Oystein A. 1995. “Referentiality and Argument Positions in Icelandic.” Working Papers in Scandinavian Syntax 55:89-109. Department of Scandinavian Languages, University of Lund. ——. Forthcoming. “A Configurational Approach to Noun Phrase Interpretation.” In Joao Costa, Rob Goedemans, and Ruben van de Vijver, eds., Proceedings of ConSOLE IV. Holland Academic Graphics, The Hague. Van Valin, Robert D. 1990. “Layered Syntax in Role and Reference Grammar.” In Jan A. Nuyts, Machtelt Bolkenstein, and Co Vet, eds., Layers and Levels of Representation in Language Theory. John Benjamins, Amsterdam, pp. 193-232. Vegnaduzzo, Milena. 1997. Analisi semantica e sintattica dell’avverbio italiano “ancora”. Tesi di laurea. University of Padua. Vendler, Zeno. 1967. Linguistics in Philosophy, Cornell University Press, Ithaca, N.Y. —_——. 1984. “Adverbs of Action.” In Papers from the Parasession on Lexical Semantics. Chicago Linguistic Society, Chicago, pp. 297~307. Venier, Federica. 1991. La modalizzazione assertiva: avverbi modali e parentetici. Franco Angeli, Milano. Verkuyl, Henk. 1972. On the Compositional Nature of the Aspects. Reidel, Dordrecht. Vikner, Carl. 1978. “Les auxiliaires négatifs: fonction et position.” Revue Romane 13:88109.

—— —. 1986. “Aspect in French: The Modification of Aktionsart.” In Finn Sgrensen, ed., Aspects of Aspect: Three Studies on Aspect in English, French and Russian. Nyt Nordisk Forlag Arnold Busck, Copenhagen, pp. 58-101. Vikner, Sten. 1985. “Reichenbach Revisited: One, Two, or Three Temporal Relations.” Acta Linguistica Hafniensia 19:81—-98. ——. 1988. “Modals in Danish and Event Expressions.” Working Papers in Scandinavian Syntax, no. 39. Department of Scandinavian Languages, University of Lund. ———. 1991. “Relative der and Other C° Elements in Danish.” Lingua 84:109-—136. ————., 1994, “Scandinavian Object Shift and West Germanic Scrambling.” In Norbert Corver

and Henk van Riemsdijk, eds., Studies on Scrambling. Mouton de Gruyter, Berlin, pp. 487-517.

256 REFERENCES

—-——. 1995. Verb Movement and Expletive Subjects in the Germanic Languages. Oxford University Press, New York. Vinet, Marie-Therése. 1995. “Adverbial Quantifiers, Negation and Stress Rules Effects.” University of Venice Working Papers in Linguistics 5(1):115-138. ——__—. 1996. “Adverbial Quantifiers and Dialect Variation in a Minimalist Framework.” In James R. Black and Virginia Motapanyane, eds., Microparametric Syntax and Dialect Variation. John Benjamins, Amsterdam, pp. 213-227. Vonen, Arnfinn Muruvik. 1994. “The Expression of Temporal and Aspectual Relations in Tokelau Narratives.” In Carl Bache, Hans Basbgll, and Carl-Erik Lindberg, eds., Tense Aspect and Action: Empirical and Theoretical Contributions to Language Typology. Mouton de Gruyter, Berlin, pp. 371-396. von Wartburg, Walther, and Paul Zumthor. 1958. Précis de syntaxe du francais contemporain. Editions A. Francke S.A., Berne. Watanabe, Akira. 1993. “Larsonian CP Recursion, Factive Complements and Selection.” Proceedings of the 23rd North Eastern Linguistic Society. Department of Linguistics, University of Massachusetts at Amherst, pp. 523-537. Watson, Saundra K. 1966. “Verbal Affixation in Pacéh”, Mon-Khmer Studies, 2:15-30. Watters, James K. 1988. “Topics in Tepehua Grammar.” Ph.D. diss., University of California at Berkeley (distributed by UMI Dissertation Services, Ann Arbor, Mich.). Waugh, Linda R. 1982. “Marked and Unmarked: A Choice between Unequals in Semiotic Structure.” Semiotica 38:299-318. Weber, David J. 1989. A Grammar of Huallaga (Hudnuco) Quechua. University of California Press, Berkeley. Wee, Lionel. 1995. “Verbal Prefixation in Malay: Reconfiguring Paradigmatic Relations.” Proceedings of the 21st Annual Meeting of the Berkeley Linguistics Society. Linguistics Society, Berkeley, Calif., pp. 313-323. Weimer, Harry. 1972. “Yareba Verb Morphology.” Te Reo (Journal of the Linguistic Society of New Zealand) 15:58-73. Whitehead, Carl R. 1991. “Tense, Aspect, Mood and Modality in Menya.” Papers in Papuan Linguistics 1:245~311. Pacific Linguistics, Series A, no. 73. Australian National University, Canberra. Whitley, Stanley M. 1975. “Dialect Syntax: Plurals and Modals in Southern American.” Linguistics 161:89-108.

Wiering, Elisabeth. 1974. “The Indicative Verb in Doowaayaayo.” Linguistics 124:3355.

Wilkendorf, Patricia. 1991. “Le systéme temporel et aspectuel de la langue nomadnde.” In Stephen Anderson and Bernard Comrie, eds., Tense and Aspect in Eight Languages of Cameroon, Summer Institute of Linguistics and the University of Texas at Arlington, pp. 105-146. Willett, Thomas. 1988. “A Cross-Linguistic Survey of the Grammaticization of Evidentiality.” Studies in Language 12:51—97. Winford, Donald. 1998a. “Tense and Aspect in Sranan and the creole prototype’, ms., Ohio State University. ————— 1998b. “Irrealis in Sranan. Mood and Modality in a radical] creole”, ms., Ohio state University.

Wise, Mary R. 1986. “Grammatical Characteristics of Preandine Arawakan Languages of Peru.” In Desmond C. Derbyshire and Geoffrey K. Pullum, eds., Handbook of Amazonian Languages, vol. 1. Mouton de Gruyter, Berlin, pp. 567-642. Woisetschlaeger, Erich F. 1977. “A Semantic Theory of the English Auxiliary System.” Ph.D. diss., MIT.

REFERENCES 257

Wongbiasaj, Soranee. 1979. “On the Passive in Thai.” Studies in the Linguistic Sciences 9(1):207-216. von Wright, Georg Henrik. 1951. An Essay in Modal Logic. North Holland, Amsterdam. Wyner, Adam Zachary. 1994. “Boolean Event Lattices and Thematic Roles in the Syntax and Semantics of Adverbial Modification.” Ph.D. diss., Cornell University. Yavas, Feryal. 1980. “On the Meaning of the Tense and Aspect Markers in Turkish.” Ph. D. diss., University of Kansas (distributed by UMI Dissertation Services, Ann Arbor, Mich.). Yoon, Jeong-Me. 1990. “Verb Movement and Structure of IP in Korean.” Language Research 26:343-—371.

Zagona, Karen. 1988. Verb Phrase Syntax: A Parametric Study of English and Spanish. Kluwer, Dordrecht. ——_——, 1995. “Temporal Agreement Structure: Configurational Elements of Construal.” In

Piermarco Bertinetto et al., eds., Temporal Reference, Aspect and Actionality, vol. 1. Rosenberg and Sellier, Turin, pp. 397-410. Zamparelli, Roberto. 1995. “Layers in the Determiner Phrase.” Ph.D. diss., University of Rochester. Zanuttini, Raffaella. 1997. Negation and Clausal Structure. A Comparative Study of Romance Languages. Oxford University Press, New York. Zubizarreta, Maria Luisa. 1982. “On the Relationship of the Lexicon to Syntax.” Ph.D. diss., MIT. ———. 1986. Levels of Representation in the Lexicon and in the Syntax. Foris, Dordrecht. Zwart, Jan Wouter. 1993. “Dutch Syntax: A Minimalist Approach.” Ph.D. diss., University of Groningen.

This page intentionally left blank

LANGUAGE INDEX

Abkhaz, 72, 155 Boro, 105, 106, 212n74, 213n79

Ainu, 100, 211n68 Bosnian/Serbo-Croatian, 36f, 182n95,

Akha, 201n21 182n96, 182n97, 182n98

Akuapem, 197n62 Brazilian Arawakan, 201n21 Albanian, 41f, 188n13, 201n21 Brazilian Portuguese, 211n71, 220n36 Aleut, 105, 158, 195n50, 203n30, 212n74, Bulgarian, 199n13

213n79, 229n7 Burmese, 157 American Sign Language, 187n11, 212n77 Canela-Crah6, 162, 230n15 Amuesha, 206n46 Cariri, 222n50

Anejom, 72, 160, 190n27 Catalan, 121, 146, 198n4, 227n9 Arabic, 156, 195n51, 223n52 Central Alaskan Yup’ik, 70, 158

Atchin, 207n51 Chaha, 195n51

Axininca Campa, 212n74 Chemehuevi, 213n79 ChiBemba, 70

Babele, 125 Chinese, 39-41, 55, 100, 101, 157,

Bagandji, 196n56 183n106, 183n107, 183n108, 183n109, Banggi, 74, 207n51, 210n64 183n110, 187n9, 201n21, 206n47,

Bangwa, 72f, 125 209n60

Basque, 54, 65, 66, 74, 165, 188n13, Choctaw, 100

189n20, 196n59, 201n20 Coahuilteco, 105, 165, 230n20

Bellunese, 134, 146, 147, 227n9, 227n10 Comanche, 197n62

Berber, 137, 223n52, 225n19 Cupejio, 203n34 Berbice Dutch Creole, 73, 164, 230n18

Big Nambas, 160, 209n59, 229n10 Da’a, 106

Blackfoot, 209n63 Dagaare, 156, 196n59 259

260 LANGUAGE INDEX

Danish, 79f, 85, 115, 198n5, 199n13, Guyanese Creole, 59f, 72, 74, 80f, 164,

216nb2 189n23, 189n24, 189n25, 190n26, Dghwede, 208n57 196n61, 230n17 Dieguefio, 162, 200n19

Doowaaydayo, 206n46 Haitian Creole, 61-63, 75, 96, 164,

Dschang, 201n28 189n23, 190n26, 192n35, 192n36, Dutch, 12, 18, 88, 137f, 198n6, 199n13, 192n37, 208n55

225n20 Hawaiian, 199n13

Dyrbal, 212n74 Hawick Scots. See Scots English Hebrew, 38f, 182n100, 182n101, 182n102,

Eastern Pomo. See Pomo 182n103, 182n104

Edo, 210n64 Hidatsa, 66, 88, 162, 195n54 English, 17, 19, 33f, 54, 57, 86, 89f, 91, 93, Hindi, 57, 154, 206n47 94, 95, 97, 99, 100, 103, 109, 112f, 126, Hixkaryana, 71f, 95

131, 138, 139, 153, 171n22, 174n37, Hmong, 209n63 174n38, 175n39, 178n57, 180n76, Hungarian, 100, 101, 154, 225n25

190n27, 203n31, 204n36, 205n41, Hunzib, 210n64 212n73, 213n80, 222n46, 222n47, 224n7,

224n9, 224n16, 225n22, 225n24 Ibibio, 70

Eskimo, 213n79 Icelandic, 112, 215n9, 216n12, 217n18 Evenki, 106, 154, 186n7, 212n74 Ika, 105, 163, 212n76

Ewe, 95 Inuit, 192n34

Irish, 193n42, 201n20

Faroese, 218n18 Isekiri, 74, 156, 229n4

Finnish, 154, 221n45, 223n52 Italian, 4—16, 45-47, 49-51, 79f, 84, 87~-

Fongbe, 63, 73, 191n30, 192n38, 193n40 88, 89, 90, 91, 92, 93, 95, 97, 98, 99,

Fore, 161 100, 101, 102f, 104, 110f, 113, 117f, French, 4-13, 17f, 19, 23, 46, 96, 116-120, 119, 121, 122, 123, 124, 126, 144-152, 121, 133f, 143f, 146, 168n5, 168n7, 167n4, 168n6, 168n8, 168n10, 169n11,

168n10, 169n13, 169n14, 170n18, 172n26, 172n27, 173n30, 174n37, 170n19, 171n23, 172n25, 172n26, 175n41, 177n53, 178n63, 179n71, 173n28, 173n29, 173n31, 176n46, 180n79, 180n80, 180n84, 181n85, 176n47, 176n48, 176n49, 177n53, 192n36, 193n41, 198n4, 198n5, 199n13, 178n59, 199n13, 204n36, 208n56, 203n30, 205n39, 205n40, 207n52, 211n71, 214n7, 218n21, 218n23, 207n53, 208n56, 208n57, 211n69, 218n24, 219n30, 219n33, 219n34, 211n70, 211n71, 211n72, 213n80, 220n36, 221n45, 226n1, 226n2, 228n16 213n3, 214n7, 218n25, 218n26, 218n27,

Friulian, 147, 227n10 218n28, 219n32, 223n52, 226n25, Fula/Fulfulde, 76, 100, 104, 105, 106, 156, 226n3, 228n12, 228n14, 228n15,

212n74, 230n20 228n17

Garo, 72, 157, 186n7, 207n50, 213n79 Jacaltec, 196n56 German, 12, 58, 85f, 113f, 171n19, Jamaican Creole, 201n27 176n5S0, 199n13, 206n45, 216n14, Japanese, 188n13, 206n47 217n15, 218n19, 221n45

Greek (Modern), 154, 211n71 Kachin, 65, 157 Greenlandic Eskimo, 209n58 Kako, 156, 212n78, 229n5 Gugu-Yalanji, 207n51 Kammu, 73, 74, 158, 196n57, 209n61, Gungbe, 63-65, 75, 99, 138, 156, 191n30, 229n3, 229n8 192n38, 192n39, 193n40, 228n3, 230n17 Kana, [90n27

LANGUAGE INDEX 261

Karen, 207n50 Ndyuka, 73, 164, 191n33 Kayardild, 210n64 Ngiyambaa, 84f, 161 Kewa, 56 Nigerian Pidgin, 125, 164, 230n19 Kiribatese, 160 Nkore-Kiga, 206n46, 206n50

Kiwai, 200n19 Nomaande, 70

Kobon, 99, 209n62 Northern Pomo. See Pomo Kom, 156, 229n6 Northern Sotho, 72f, 157, 195n55 Korean, 53f, 67, 71f, 84f, 86, 154, 186n3, Norwegian, 34-36, 112f, 115, 116,

187n8, 188n13, 194n43 169n12, 171n19, 175n42, 181n91,

Koyo, 190n27 181n92, 182n93, 215n7,215n9, 216n12, Koyukon Athapaskan, 194n47 217n15, 217n18, 223n52 Kristang, 74, 164

Kusaiean, 191n33 Oksapmin, 186n7, 212n78 Kutenai, 200n19

Kwaio, 97, 159, 229n3, 229n10 Pacoh, 106

Kwakiutl, 200n19 Paduan, 121, 122, 147, 220n40, 220n41, 227n10

Latin, 197n63, 199n10 Patani, 157 Lewo, 125 Pavese, 121, 147, 171n19, 220n37, 227n10 Lezgian, 72, 138, 155, 190n27 Pawnee, 74, 105, 200n19, 204n37

Lhasa Tibetan, 203n34 Piedmontese, 121, 147, 169n10, 191n33,

Lisu, 55f 220n36, 223n52, 227n10 Louisiana Creole, 164, 189n23, 196n57, Piro, 201n21

Luganda, 206n46, 206n50 Polish, 101

Pomo: Eastern, 209n58; Northern, 210n64

Macushi, 105, 163 Ponapean, 159f

Makaa, 157 Portuguese, 74, 96 Makah, 128

Malagasy, 42f Quebec French, 169n10 Malay, 74f, 159, 196n60, 204n34 Quechua: Equadorian, 56; Huallaga, 100; Malayalam, 72, 125, 158, 203n30, 229n7 Huanuco, 187n10; Imbabura, 163,

Mam, 96, 201n27 187n10, 230n17 Mandan, 210n64, 212n74 Quileute, 209n58 Maori, 102

Maranungku, 65 Rapanui, 91

Menomini, 85 Romanian, 147, 152, 227n11 Menya, 161

Milanese, 121, 147, 171n19, 220n37, Samoan, 73, 160, 190n27, 230n12

227n10 Sanio-Hiowe, 66, 161, 196n59

Mizo, 65 Sardinian, 46, 133, 146

Mofu-Gudur, 75, 156, 209n63, 228n3 Scots English, 54, 79, 196n57, 198n2

Mokilese, 209n60 Scottish Gaelic, 192n39, 193n42

Mongo, 105 Sea Island Creole, 75, 190n26

154, 228n1 Croatian

Mongolian, 72, 125, 190n27; (Khalkha), Serbo-Croatian. See Bosnian/SerboSeychelles Creole, 75, 83, 164, 192n37 Nahuatl, 74 Shona, 194n49

Nankina, 197n61 Sierra Leone Krio, 164 Navajo, 68f, 162, 194n45, 194n46, 194n47, Sinhala, 204n34

209n60 Slave, 194n47

262 LANGUAGE INDEX

Sobei, 92 Ubykh, 155

Spanish, 57, 147f, 152, 153, 214n6, Ulithian, 105

227n11 Una, 54, 96, 97, 162, 195n54 Spokane, 105 Ute, 56, 128, 163 Sranan, 60f, 165, 190n26, 190n28, 190n29,

192n33 Valdotain, 121

Swedish, 115f, 213n80, 216n12, 217n17 Venetian, 147, 227n10

Swedish Sign Language, 187n11 Vietnamese, 210n64

Tagalog, 208n58 Wahgei, 162 Tamil, 206n47 Walmadyari, 161 Tarascan, 200n19 Waorani, 163 Tauya, 161, 186n7, 195n54, 210n64, Wapishana, 208n58

212n77 Warao, 206n46 Temne, 193n42 Wayampi, 105, 210n67

Tepehua, 96, 106, 210n67 Welsh, 66f, 153, 222n50 Thai, 75, 159, 209n63, 229n9 West Flemish, 219n29

Thargari, 88 West Greenlandic, 94, 95, 188n15, 212n74, Tigak, 191n33, 196n56 212n78, 213n79

Tigrinya, 203n30 Wikchamni, 212n74

Tikar, 212n78 Wintu, 128 Tokelau, 75, 160, 230n13

Trevisan, 147, 227n10 Xhosa, 209n58 Tshangla, 157

Tsova-Tush, 203n34 Yana, 213n79 Tumbuka, 202n30 Yapese, 212n78

Tinen, 105 Yareba, 91, 162

Turkish, 54, 56, 67f, 71, 72, 81, 90, 125, Yavapai, 104, 163

137, 155, 186n4, 186n5, 188n13, Yimas, 97, 194n48 190n27, 193n43, 194n43, 198n2, Yoruba, 157, 199n13 209n58, 223n52

Tuyuca, 105, 206n46, 222n51, 223n52 Zuni, 72, 165, 230n21

NAME INDEX

Aboh, Enoch, 64, 138, 185, 192n38, Beninca’, Paola, 112, 167, 183, 185,

193n40, 201n27 172n27, 191n33, 197, 198n5, 213,

Acquaviva, Paolo, 168n10, 171n20 220n40, 220n41, 221n44, 223, 223n2,

Adamson, Lilian, 61 226n25, 227n10

Afarli, Tor, 227n7 Bernstein, Michael, 189n22 Aksu-Kog, Ayhan, 56, 81, 186n4 Bertinetto, Piermarco, 91, 204n35, 207n51 Alexiadou, Artemis, 183n1, 211n71 Bickerton, Derek, 58, 60

Andrews, Avery, 25 Binnick, Robert L, 61, 95, 201n27, 209n60 Arnott, D.W., 76, 93, 104 Bisetto, Antonietta, 227n10 Bobaljik Jonathan D., 216n12

Babu, Hany, 72, 158, 229n7 Boertien, Harmon S., 54, 186n6 Baker, Carl L., 8, 132, 174n34, 177n56, Borer, Hagit, 200n17

213n4 Brown, Keith, 54, 78f, 126

Baker, Mark, 52, 68, 70, 185, 186n2, Brugger, Gerhard, 167, 171n19, 197,

187n11 217n15, 218n19

Baltin, Mark, 189n22 Burzio, Luigi, i111

Battistella, Edwin L., 54, 186n6 Bybee, Joan, 55, 70, 72, 74, 101, 128, 129,

Bayer, Josef, 30f, 169n12 135, 162, 187n7, 189n18, 195n50, Beghelli, Filippo, 108, 114 195n51, 200n19, 201n25, 201n26, Bellert, Irena, 11, 88, 203n31 204n37, 205n43, 207n50, 225n18 Belletti, Adriana, 31, 52, 143, 144, 148, 167, 170n15, 170n16, 171n20, 174n37, Calabrese, Andrea, 197n63, 218n26

178n63, 180n80, 180n81, 183, 185, Camporese, Nadia, 182n105, 182n106, 185n13, 185n14, 211n70, 226n6, 227n7 183n107, 183n109, 183n110

263

264 NAME INDEX

Cardinaletti, Anna, 7, 8, 17, 21, 167, Fillmore, Charles J., 25, 28 168n10, 173n30, 175n42, 178n64, Foley, William A., 55f, 59, 97, 135, 161, 181n89, 213, 215n8, 217n15, 218n26, 187n11, 188n14, 189n18, 194n48

219n28, 223 Fortescue, Michael D., 94, 95

Chhangte, Lalnunthangi, 65 Frawley, William, 61, 95, 99 Chierchia, Gennaro, 26, 99

Chomsky, Noam, v, 16, 20f, 25, 28f, 45, Gibson, Kean, 59, 72, 98 52, 68, 98, 102, 108, 109, 132, 133, Giorgi, Alessandra, 81, 83, 131, 133,

183n2, 184n8, 189n18 200n14, 223

Cinque, Guglielmo, 17, 22, 29, 43, 44,112, Giusti, Giuliana, 139, 167, 218n20, 218n26 139, 146, 148, 170n15, 174n32, 178n61, Givén, Talmy, 56, 70

178n63, 194n45, 210n66, 2111n70, Grimshaw, Jane, 84 211n71, 220n39, 221n45, 222n47, Groat, Erich, 212n73, 223

223n5, 226n6 Groot, Casper de, 101, 162 Cole, Peter, 163, 230n17 Comrie, Bernard, 81f, 91, 96, 98, 99, 105, Haegeman, Liliane, 180n76, 217n15,

128, 129, 196n58, 199nit, 199n12, 219n29 201n27, 204n35, 205n38, 206n46, Haiden, Martin, 108 206n48, 206n50, 208n55, 224n11, Haider, Hubert, 167, 197

229n3, 230n13 Haspelmath, Martin, 72, 95, 138, 155,

Corver, Norbert, 139 228n2

Culicover, Peter, 167, 197, 194n48 Hendrick, Randall, 66f Cunningham, Irma A.E., 54, 75 Hengeveld, Kees, 84, 162, 188n14, 200n19 Hetland, Jorunn, 177n50

Dahl, Osten, 83, 91, 97, 99, 187n11, Holmberg, Anders, 115, 154, 213n80,

190n30, 192n39, 206n46, 208n55, 216n12, 217n18, 221n45

209n58, 223n52 Hornstein, Norbert, 81, 83, 94, 199n12, Déchaine, Rose-Marie, 167, 185, 223n4 200n14 DeGraff, Michel A.F., 61, 63, 185, 189n23, Huang, James, 182n105 191n32, 192n35, 223n4

Delfitto, Denis, 204n35 Iatridou, Sabine, 197n1, 224n11 Déprez, Viviane, 168n7

Derbyshire, Desmond C., 95 Jackendoff, Ray, 8, 11, 19, 86, 89, 137,

Desouvrey, L.H., 63 174n35, 175n38, 180n79, 203n32, Diesing, Molly, 108, 216n14, 217n15, 217n18 225n24, 213n4

Dik, Simon, 59, 177n54, 188n14 Jamet, Marie Christine, 167, 170n18,

Dini, Luca, 179n71 171n19

Dobrovie-Sorin, Carmen, 147, 152, 227n11 Jonas, Dianne, 215n9, 216n12

Egerland, Verner, 115, 213, 217n1t5 Kallulli, Dalina, 183n111 Engver, Karl, 9, 143, 168n9, 173n28, Kayne, Richard S., v, 7, 10, 16f, 21, 22, 25,

173n29, 226n2 29, 31, 43, 44, 45, 57f, O5f, 67, 68, 86,

Ernst, Thomas B., 31, 174n33, 174n34, 108, 111, 112, 118, 126, 140, 148, 167,

185n13, 205n41, 221n45, 222n46 176n49, 178n60, 178n63, 179n66,

Erteshik-Shir, Naomi, 228n14 179n67, 181n88, 181n89, 183, 183n1, 185, 195n53, 197, 204n36, 211n70, Fagerli, Ole Torfinn, 71, 76, 93, 100, 104, 105 212n73, 212n77, 213, 213n4, 214n4,

Fassi Fehri, Abdelkader, 156, 182n99, 214n7, 218n23, 218n25, 219n34,

195n51, 223n52 222n46, 222n47, 223, 224n7, 224n8,

Figueiredo, Cristina, 211n71 224n16, 227n7, 228n16, 229n3

NAME INDEX 265

Kenesei, Istvan, 154 Napoli, Donna Jo, 219n28 Keyser, Samuel J., 102, 177n55 Nichols, Johanna, 85, 188n14

Kiss, Katalin E., 216n14 Nichols, Lynn, 165, 230n21

Koizumi, Masatoshi, 179n75 Nilsen, Oystein, 28, 29, 116, 167, 169n12, Koopman, Hilda, 137, 138, 191n31, 175n42, 177n50, 181n90, 181n91, 191n33, 192n36, 225n21, 225n24 181n92, 181n93, 213, 215ni1, 217n15, Kornfilt, Jaklin, 54, 68, 72, 185, 194n43, 217n18 198n2, 209n58

Koster, Jan, 12, 18, 177n52, 227n7 Obenauer, Hans G., 179n69, 205n39 Kratzer, Angelika, 201025, 216n14 Ordéfiez, Francisco, 111 Ortiz de Urbina, Jon, 66, 74, 188n13

Ladusaw, William, 174n32, 221n42 Ouhalla, Jamal, 66f, 136f, 187n11, 188n16, Laenzingler, Christopher, 177n53, 180n81, 223n52, 224n6, 225n19

183nl Oyharcabal, Bernard, 54

Laka, Itziar, 136, 189n20, 201n20, 222n46

Larson, Richard, 22, 28 Palmer, Frank R., 72, 78, 84f Leblanc, Annie, 61, 62, 63 Paoli, Sandra, 3 Lefebvre, Claire, 61, 73, 191n30, 191n31, Papen, Robert Antoine, 75

191n32, 191n33, 192n36, 192n37, Parry, Mair, 220n36

192n38 Parsons, Terence, 28

Leko, Nedzad, 96, 182n94 Pesetsky, David, 179n75, 223 Lepschy, Anna Laura, 7 Pianesi, Fabio, 81, 83, 131, 133, 200n14,

Lepschy, Giulio, 7, 223 223

Lindstedt, Jouko, 82 Picallo, Carme M., 186n6, 198n4 Lois, Ximena, 147, 152, 214n6, 226n4, Platzack, Christer, 86, 115, 185, 213n80,

227n11 216n12, 217n18, 227n7

Lonzi, Lidia, 7, 16, 28 Poletto, Cecilia, 126, 167, 171n19, 171n20,

Louwerse, J., 54, 96, 97 177051, 185, 188n16, 191n33, 197, Lyons, John, 78, 128 198n5, 218n19, 224n6, 226n25, 227n10 Pollock, Jean-Yves, 5, 17, 45, 46, 47, 52,

Magloire-Holly, Héléne, 62f 143, 167, 169n10, 170n19, 224n15,

Mahajan, Anoop, 37 226n2

Mahootian, Shahrzad, 167n3 Prinzhorn, Martin, 213 Manfredi, Victor, 167, 185 Progovac, Ljiljana, 182n94, 182n97,

Manzini, Rita, 167, 184n7 182n98

May, Robert, 222n48 Puskas, Genoveva, 225n?5 McCawley, James D., 179n70, 202n29

McCloskey, James, 108, 180n81, 201n20, Rackowski, Andrea, 43, 183n112

203n32, 222n48 Reichenbach, Hans, 59, 73, 81

Mchombo, Sam, 70 Reinhart, Tanya, 22

Michaelis, Laura A., 94, 206n46, 207n51 Rescher, Nicolas, 78, 84

Michaelis, Susanne, 164 Reuland, Eric, 216n14

Mitchell, Erika, 108, 136, 154 Rice, Keren, 68, 194n44, 194n46, 194n47

Mithun, Marianne, 70, 158 Riemsdijk, Henk van, 137, 138, 225n23 Moltmann, Friederike, 98, 108, 167, Rivero, Maria-Luisa, 154, 211n70 177n50, 178n57, 206n45, 216n14 Rizzi, Luigi, 84, 122, 145, 167, 177n51,

Moro, Andrea, 227n10 183, 185, 186n3, 201n20, 212n77,

Muller, Claude, 169n10, 170n19 218n19, 220n41, 225n25, 226n3, 226n6

Munaro, Nicola, 227n10 Roberts, fan, 12, 89, 90, 102, 179n75,

Muysken, Pieter, 56, 58 186n6, 203n32, 218n18

266 NAME INDEX

Rochette, Anne, 20, 205n41 Sufier, Margarita, 227n11 Roeper, Thomas, 102, 177n55 Svantesson, Jan-Olof, 72, 73, 74, 154, 158,

Rooth, Mats, 30f, 169n12 196n57, 209n61 Ruwet, Nicolas, 28, 176n49

Taraldsen, Tarald, 177n50, 227n7

Sadock, Jerrold N., 88 Tenny, Carol, 200n17

Santorini, Beatrice, 167n3 Thrainsson, Hoskuldur, 54, 79 Schlyter, Suzanne, 174n33, 175n38, Togeby, Knud, 5, 10, 146, 168n7, 169n14,

175n40, 176n46, 178n59, 180n84, 170n19, 171n23

212n76, 221045 Travis, Lisa, 20, 68, 89, 93, 174n38, 185,

Seuren, Peter A., 60f 203n32 Shionsky, Ur, 139, 182n99, 182n100,

182n102, 182n104, 218n20 Vai, Massimo, 197n63, 227n10

Siewierska, Anna, 101 van der Auwera, Johan, 172n24, 206n46 Sigurdsson, Halldér Armann, 112, 213, Vangsnes, Mystein A., 215n9

216n12 Van Valin, Robert D., S55f, 59, 135,

Singhapreecha, Pornsiri, 159, 209n63, 187n11, 188n14, 189n18

229n9 Vegnaduzzo, Milena, 172n24

Slobin, Dan, 56, 81 Vendler, Zeno, 12, 83, 200n15, 205n41 Smith, Carlota, 55, 100, 101, 200n15, Vikner, Carl, 9, 171n19, 199n13

209n60, 212n76 Vikner, Sten, 54, 79, 81-83, 129, 185, 197, Smith, Norval, 61 199n9, 199n12, 199n13, 200n14, 215n9, Sohn, Ho-Min, 53, 105 216n12, 217n18, 225n25, 227n7 Speas, Margaret, 68, 187n11, 194n45 Vinet, Marie-Therése, 167, 219n31 Spears, Arthur K., 61

Spellmire, Susan, 79, 198n3 Xu, Liejiong, 182n105, 182n108 Sportiche, Dominique, 7, 44, 109, 116,

117, 118, 167n1, 178n59, 218n21, Zagona, Karen, 83, 189n18, 200n14

218n23, 218n24, 223n5 Zanuttini, Raffaella, 3, 120, 121, 122, 125, Starke, Michal, 7, 8, 17, 21, 167, 175n42, 167, 169n10, 219n35, 220n36, 220n38,

178n64, 181n89, 223n5 220n41, 224n15, 226n5, 227n10 Stowell, Tim, 108, 114, 199n12, 200n14 Zubizarreta, Maria Luisa, 89

Sueur, Jean-Pierre, 11f, 18 Zwart, Jan Wouter, 58, 185

SUBJECT INDEX

A-bar operator position, 19, 109 Adverb(s)

Ability, 79-81, 90, 203n33 affirmative, 171n20 abitualmente, 8, 169n12, 204n36 agent-oriented, 203n32

acabar de, 96 Albanian, 41f

Accomplishments, 83, 200n15, 200n17, alethic, 203n31

209n60 Bosnian/Serbo-Croatian, 36f

Achievements, 83, 99, 200n15, 209n60 Chinese, 39--41 Activities, 83, 101, 200n15, 200n17, Circumstantial (see Adverbial(s))

209n60 climbing, 17f, 109 Addressee agreement, 54 completive aspect, 209n60 Adjunction, 44, 46, 47, 50, 68, 108, 186n2, Criterion, 183n1 194n45 deontic modal, 90 Admirative. See Mood domain, 11, 13, 168n5, 174n33, 175n41 Adverbial(s) durative, 98, 200n15 bare DP, 202n29 English, 33f

circumstantial, 28-30, 87, 98, 208n56 epistemic, 18, 84, 86, 88, 135, 174n34,

durative (see measure) 174n37, 174n38, 203n31 frame, 174n33 evaluative, 11-13, 18, 33, 84f, 174n34,

locative, 16 174n37, 174n38, 203n30, 217n15 manner, 15f, 28, 224n10 (see also evidential, 85f, 135, 174n37, 203n30 Adverb(s)) focusing, 30-32, 168n5, 169n12,

measure, 98, 208n58 169n13, 170n18, 180n79, 180n80,

place, 15, 191n32 207n52, 213n78, 214n4, 226n4 purpose, 15 French, 4—13 temporal, 15f frequentative, 26—28,90, 91-93, 103, 104 267

2068 SUBJECT INDEX

Adverb(s) (continued) subcategorized, 17, 19 habitual, 4f, 14, 90f, 136, 168n5, subject-oriented, 11—13, 19, 89f, 174n35, 168n10, 171n20, 180n80, 204n37 174n38, 175n39, 176n45, 203n32,

Hebrew, 38f 205n41, 216n11

“higher”, 11-13, 14, 16, 31, 32, 49f, temporal, 12, 15, 87f 145, 146, 148f, 180n8, 218n19, terminative aspect, 136

22707 of time (see temporal)

inchoative, 212n76 viewpoint, 174n33

incorporation of, 211n70 Agreement, 53, 108

Italian, 4-16 affatto, 8,171n20

iterative, 69, 91, Affix

“lower”, 4-11, 13-16, 32, 33, 113, 149f, emphatic, 200n19 152, 182n105, 184n7, 185n14, 214n7, (non)volitional, 203n34

220n36, 226n1 again, 174n32

Malagasy, 42f Alethic. See Modal(s) of manner, 7, 17, 19f, 102, 174n38, allerede, 169n12

191n32, 205n41 (see also almost, 99

Adverbial(s)) almost-aspect. See Aspect modal, 11-13, 18, 174n34, 174n35, already, 94, 171n22, 206n46

174n37, 175n39 altro, 206n49

movement of, 16-19, 178n58, 178n59 always, 33

negative, 5, 8, 171n20, 182n93 A-movement, 222n48

Norwegian, 34—36 ancora

parenthetical, 32, 174n36 continuative (‘still’), 32, 95f, 172n24, place, time, manner (see Adverbial(s)) 172n26, 174n32, 207n52, 208n54 pragmatic, 11-13, 174n38, 175n39 (see non (‘not yet’), 8-10, 32, 172n23,

also speech act) 174n32, 185n14

presuppositional, 174n32, 217n15 repetitive (‘again’), 32, 172n24, 172n26,

prospective aspect, 135f 174n32

proximative aspect, 135f Anterior, 53, 59, 61, 73f, 75, 82f, 94, 96, quantificational, 26, 92, 169n12, 181n89, 195n55, 196n59, 199n11, 199ni14,

204n36, 204n37, 205039, 205n40, 208n55, 229n3

207n53 Anterior of the future. See Future perfect

realis mood, 180n80 Anterior of the past, 164, 196n59, 229n7.

repetitive, 90, 91-93, 104 See also Pluperfect.

retrospective aspect, 135f Antisymmetric (theory), 25, 43, 44,

reversative aspect, 69, 220n39, 225n19

scope of, 16-28 any longer, 33. See also no longer semeliterative, 69 anymore, 174n32, 181n89 sentence, 11-13, 14, 16, 18, 175n41, AP, 43, 139 180n81, 202n30 appena, 97, 100, 125, 208n56

of setting, 15, 28, 87, 174n33, 202n30, Aspect

208n56 ‘almost’, 192n39, 210n64 (see also

speaker-oriented, 11, 86, 174n38, prospective; ‘unrealized’)

speech act, 12, 33, 42f, 84f, 175n40, “capacitatif’, 105 (see also success; 181n91, £82n100, 182n102, 183n112, frustrative) 202n30, 205n41, 217n15, 228n12 celerative, 76, 93, 103f, 129f, 156,

speech time, 11-13, 15, 181n91, 206n44, 212n/74, 212n75

228n12 completive, 75f, 83, 100f, 102, 104, strong, 181n89 129f, 206n47, 210n66

SUBJECT INDEX 269

conative, 105, 212n77 semelrepetitive (see repetitive) continuative, 95f, 129f, 172n24, 207n51, success, 105, 212n77 (see also frustrative)

207n52, 212n77, 228n2, 229n5 ‘suddenly’, 208n58

continuous, 96 terminative, 94f, 96, 129f, 157, 172n24,

delayed, 105 207n52, 212n77

delimitative, 105 ‘unrealized’, 197n62 durative, 59, 76, 96, 98, 105, 129f, usitative (occasionally), 105, 204n37

187n11, 189n25, 196n61 Aspectual verb(s), 95, 212n77

‘finally’, 163 Auxiliary, 57, 58, 131f, 147, 188n16, frequentative, 104, 105, 129f, 187n11 188n17, 189n22, 224n6, 227n11

frustrative, 105 completive, 100, 211n68

generic, 99, 129f future, 201n21

grammatical, 83, 205n38 prospective, 210n64 habitual, 59, 61, 64, 70, 73f, 90f, 129f, raising, 19,177n56

189n25, 195n50, 196n61, 212n77, ‘still , 207n50 229n5 ‘suddenly’, 209n58 inceptive, 70, 105, 191n30, 196n58, -to-COMP, 122, 145, 148f, 227n7

212n76, 212n77, 229n6 ‘yer’, 207n50

inchoative, 187n11, 212n76

ingressive (see inceptive) beaucoup, 11, 171n19, 173n31 iterative, 70, 76, 91, 204n38, 230n18 ben, 171n20, 21170

(see also frequentative) bene, 7f, 10f, 46, 101-103, 119, 170n16, (im)perfect, 57, 66, 70, 73-75, 91, 173n31, 211n70, 211n71, 214n7,

94f, 96, 129f, 184n8, 187n10, 228n17

187n11, 193n40, 196n59, 206n47, bien, 7f, 118, 146, 171n20 207n51, 212n77, 229n3, 229n4, brevemente, 98, 100, 125 229n6

lexical, 83, 129f, 205n38 can, 78f, 132

perfective (see completive) Caucasian, 139 pofective (see delimitative) Cessative. See Aspect terminative predispositional, 187n11, 212n77 Checking, 131, 186n2, 195n51

pretensive, 106, 165 Clefting, 31

progressive, 59, 61, 64, 66, 67, 68, 70, Clitic Climbing, 212n77 74, 75f, 83, 90, 99, 129f, 179n71, Clitic Left Dislocated Phrase, 145f,

187n10, 189n25, 191n30, 196n61, 227n8 209n63, 210n66, 212n76, 228n3, Clitic voice, 197n63

229n4, 229n5, 229n11 Code switching, 167n3 prospective, 75, 99, 129, 132, 191n30, Complementizer, 84, 222n50

192n39, 199n10, 209n62, 209n63, negative, 201n20

210n64, 210n66, 228n3 prepositional, 146, 224n8, 226n6 proximative, 96-98, 129f, 192n34, tensed, 201n20

209n59, 212n76 completamente, 7f, 21f, 100f, 103, 104,

repetitive, 76, 77, 104, 91-93, 129f, 165, 119f, 170n16, 170n17, 170n18,

172n26, 204n38 210n66, 228n17

retardative, 212n74 completely, 34, 100f, 178n57 retrospective, 61, 75f, 96-98, 129f, complétement, 7f, 119, 170n18, 170n19 190n30, 208n55, 208n56, 210n66, Completion, 100f, 187n9

224n17 Computational system, 135

semelfactive, 205n38 Conditional, 59, 62, 64, 199n12

semeliterative, 205n38 could, 78f, 132

270 SUBJECT INDEX

CP, 18 Finite(P), 210n20, 218n19, 225n25 split, 177n51 Focalized phrases, 84, 145

wh-exclamative, 226n25 Focus movement, 31, 50 Creole languages, 58—63, 189n23, 189n24, = FocusP, 225n25

189n25, 190n26, 190n29, 190n30, Force(P), 84, 218n19, 225n25 191n31, 191n32, 191n33, 192n35, forse, 12f, 87f, 89, 174n36, 177n53

192n37 frequentemente, 169n12

Crossing, 23, 117 frequently, 179n70

Cycle, 186n2 Future, 59, 61f, 64, 72f, 79, 82f, 87f, 89, 97, 135, 199n14, 201n21, 201n28,

Default, 83, 88, 103, 127, 128-132, 133, 209n63 202n30, 211n71, 224n13. See also certainty (see realis)

unmarked irrealis, 62, 164, 230n18 déja, 5~7, 168n7 participle, 199n10

Demonstratives, 139 of the past, 59, 62, 72f, 190n27, 199n12, Deontic, 62, 90, 192n35, 198n5, 198n6, 201n23

221n46. See also adverb(s) perfect of the past, 82, 199n12

Derivation, 70, realis, 62, 163

di gia, 14, 21f, 168n6 généralement, 4f, 168n5 di nuovo, 169n12, 221n44. Generic di solito, 8, 204n36, 207n53, 221n44

DP operator, 26, 99 -movement, 109, 214n5 sentence, 91, 99, 103, 223n4

object, 109, 115f Germanic, 18, 33, 110, 128, 137f, 139,

quantified, 113f, 222n48 225n25, 227n7 -related functional projections, 108—120 Gerund, 151f

subject, 109, 110-115 gia, 5~9, 17, 92, 93, 94, 103, 120, 169n12, 169n13, 170n16, 171n20, 172n24,

early, 103f 174n32, 206n46, 221n44 encore, 9f, 171n23, 172n25, 172n26 guére, 170n19, 220n37 enda ikke, 181n92

Evaluative. See Adverb(s); Modal; Mood; Head-final languages, 57, 58, 65f, 138,

Particle; Suffix 194n45

Evidential. See Adverb; Mood; Particle; Head-initial languages, 57, 58, 65f, 140

Suffix helaas, 12, 18

évidemment, 177n53 heureusement, 12,18f, 177053

Expletive, 11 honestly, 205n41 null, i11f, 215n9

Extended projection, 84, 137, 224n8 ikke, 171n19, 175n42, 181n92, 182n93, 217n18

fast, 103f, 212n73, 224n9 ikke enda, 181n92

Feature ikke lenger, 35, 112, 182n93, 217n18 anterior, 103 Illocutionary force, 84, 128, 200n18 generic, 103 immediatamente, 97f negative, 222n47 ‘immediately’ affix, 208n57 passive, L02f imminently, 99, 135 perfect, 102f improvvisamente, 208n58 progressive, 103 Inceptive. See Aspect

weak/strong, 211n71 Incorporation, 68-70, 138, 190n27,

finally, 105 194n48, 198n3, 211n70, 225n22

SUBJECT INDEX 271

Infinitives double, 54, 198n3, 198n6, 198n7 French, 5f, 9f, 143f, 173n29, 226n1, epistemic, 53, 54, 62, 78-81, 86f, 124,

226n2 128, 135, 198n3, 198n4, 198n5,

Italian, 144-146, 226n3, 226n4, 226n5, 198n6, 203n31, 221n46,

226n6, 227n7, 227n8 evaluative, 84 intentionally, 25-28, 90, potential, 81, 155

Spanish, 226n4 necessitative, 81, 161 root, 54, 55, 60, 78-81, 89, 129f,

ja doch, 113f, 217n15 190n28, 201n25 jamais, 9f speaker-oriented, 201n25 plus, 9f, 173n29 subject-oriented, 79, 20in25

just, 96-98, 135f suffix, 198n2 triple, 79

lately, 208n56 volitional, 129f, 165, 223n3

leider, 12 molto, 11, 173n31

LF, 108, 114f, 133, 221n45 Mood(s), 78

loro, 117f, 218n24, 218n26, 218n27, 219n32 admirative, 201n21

lungamente, 98 conditional, 86, 201n23 declarative, 53, 78, 84,

mai, 9f, 170n16, 228n15 evaluative, 53, 78, 85, 128 pitt, 9, 170n16, 172n27 evidential, 53, 55, 78, 85f, 201n22,

male, 7f 201n23, 128 mal, 7£, 176n49 illocutionary, 200n19 malheureusement, 177n53 imperative, 84

Marked, 83, 91, 95, 102f, 126, 127, 128— indicative, 78

132, 133, 211n71, 223n1, 223n2, interrogative, 53, 78, 84

223n3 irrealis, 55, 66, 73, 78, 88f, 129, 196n56, meramente, 180n79 ‘perhaps’ , 88

may, 78f 196n57, 226n3, 230n12 mia (Pavese), 121, 171n19 potential, 221n45 mica, 4—6, 47, 50f, 120, 121, 126, 152, probable, 186n7

167n4, 168n6, 168n7, 168n8, 168n10, realis, 55, 78, 129, 180n80 170n16, 171n19, 171n20, 172n23, speech act, 53, 55, 78, 84, 186n3,

174n32, 175n42, 184n7, 184n10, 200n18

184n11, 184n12, 184n13, 214n6, subjunctive, 78 214n7, 216n13, 221n44, 228n14 must, 79, 86 Middle(s). See Voice

might, 79, 186n6 neanche, 8

minga (Milanese), 121, 171n19 neancora, 14 Mirative. See Mood admirative necessariamente, 89, 221n44, 226n3

Mirror Principle, 52f, 54f, 56, 66, 68, 70, Necessity

104, 186n2, 194n45 alethic, 79-81, 198n3, 198n5

misschien, 88 conjectural, 198n5

Modal(s), 78-81 Negation, 120, 126, 198n3 agent-oriented, 201n25 phrase(s), 120-126

alethic, 78-81, 89, 129f, 197n1, 198n3, presuppositional, 121, 167n4, 220n38

198n4, 198n5, 198n6, 198n7 preverbal, 121-124, 220n39, 220n40, deontic, 62, 90, 192n35, 198n5, 198n6, 220n41, 221n44, 221n45, 222n50

221n46 postverbal, 121, 220n36, 220n37, desiderative, 230n20 220n38

272 SUBJECT INDEX

Negation (continued) completive aspect, 100f, 156 real, 121, 220n38, 220n39, 220n41 continuative aspect, 156, 207n51,

scope of, 121-124, 221n45, 222n48, 229n5

222n49, 223n53 delayed aspect (‘finally’), 105 sentence, 109 durative aspect, 59, 189n25, 196n61

nemmeno, 8 evaluative, 84, 201n21 nen (Piedmontese), 121 evidential, 66, 86, 162, 188n13 neppure, 8 frequentative aspect, 157, 162

Nesting, 117 functional, 58—66, 189n22

no 230n19

nicht, 171n19 future tense, 59-66, 72f, 97, L59f, 164, niente, 10f, 173n30 209n63, 229n4, 229n9, 230n18, Italian, 8, 171n20 habitual aspect, 59, 159f, 162, 164, 165,

Milanese, 171n19 189n25, 209n61, 229n4 Paduan, 121 inceptive aspect, 156 Pavese, 171n19 inchoative aspect, 162

no longer, 94, 95, 181n89. See also any interrogative, 66, L56f, 162

longer irrealis mood, 158, 160, 164

no more, 181n89., See also any more modal, 66, 189n20

non (Italian), 121, 124 mood, 61, 189n20 Nonconcatenative morphology, 195n51 passive, 159

NP, 139 past tense, 59-66, 72f, 156f, 160, 162,

normalmente, 8 164, 230n19

n’t, 126, 222n47 perfect aspect, 66, 74, 156, 158, 160, nulla. See niente 164, 190n27, 196n60, 229n4, 229n8 priorative, 212n78 (see also Suffix,

Object Shift, 115, 116, 217n18 priorative)

Obligation, 79-81, 90, 203n33, 205n42. progressive aspect, 59, 61, 64, 65, 67,

See also Deontic 79, 156f, 159f, 162, 164, 190n27, obviously, 201n24 193n39, 209n63, 228n3, 229n4, often, 26, 180n84, 204n36, 205n39, 229n9, 230n19

205n42 prospective aspect, 64, 75, 159, 160, 164, 209n63, 228n3, 229n9

pa proximative aspect, 159, 229n10

Piedmontese, 121 retrospective aspect, 61, 75, 96, 164,

Valdotain, 121 229n10 pa pi (Piedmontese), 169n10 terminative aspect, 160 Participle partiellement, 8 absolute past, 148-150 parzialmente, 8, 125, 170n17

pa nen (Piedmontese), 220n38 tense, 59-66, 72f, 97, 209n63

active (past) , 45-49, 102f, 146-148, pas 170n16, 211n70, 211n71, 224n8, Catalan, 167n4, 168n5, 227n9

22/nil French, 4f, 120, 168n5, 168n7, 168n9,

211n71 176n47 present, 150f Passive. See Voice

passive, 102f, 147f, 171n19, 211n70, 168n10, 169n14, 171n19, 171n23,

Particle(s) Past, 53, 59, 72, 79, 82f, 87f, 89, 135, anterior tense, 59, 74f, 156f, 164, 201n28, 229n7

196n60, 228n3, 230n18, 230n19 hesternal, 208n55 aspect(ual), 59, 61, 74f, 189n20, 191n30 hodiernal, 208n55

SUBJECT INDEX 273

immediate, 190n30, 208n55, 208n58 probablement, 18, 177n53

irrealis, 191n33 probably, 11, 124, 174n37, 174n38, perfect, 60 (see also Anterior of the past 176n45, 181n88, 214n4, 221n46

and Pluperfect) Progressive. See Aspect

perfino, 180n79, 180n80, 181n85 pur, 171n20 perhaps, 11, 88, 175n39

Permission, 79—81, 90, 203n33 Quantification

peu, 11, 171n19 at a distance, 205n39 peutétre, 18, 177n53 over events, 26, 92, 93, 103, 169n12, PF, 213n2 204n36, 205n39, 205n40, 205n44, piu, 5, 9, 13f, 47, 94f, 120, 168n8, 168n10, 207n53, 224n9 170n16, 172n24, 174n32, 206n49, over processes (or states), 26, 92, 93,

228n15 103, 169n12, 204n36, 205n39

mai, 9, 172n27, 173n29 Quantifier

Pluperfect, 53, 196n59 (see also Anterior floating, 116-120, 218n23, 218n24,

of the past) 218n25, 218n26, 218027, 219n28,

plus, Sf, 9, 119f, 168n10, 171n19, 172n23, 219n30, 219n32, 219n33, 219n34

173n28, 173n29, 176n48, 219n33 raising, 114 jamais, 9, 173n28, 173n29 quasi, 99, 100, 104, 125, 210n65, 221n44

poco, 11, 173n31 quickly, 93, 103f, 206n44, 206n45, 224n9

poi, 8f, 171n21, 172n26 Quotative, 85, Polarity

items, 222n46 rapidamente, 93, 104 phrase, 126 raramente, 205n40, 207n53

Politeness recentemente, 97, 208n56 functional projection, 224n1 1 recently, 97

suffix, 54 Reconstruction, 17, 25, 221n45 Possibility, alethic, 79-81, 198n3, 198n4, Reduplication, 207n51 198n6 Reference time, 81f, 94, 96, 97, 199n9 PP, 137~139 Relativized Minimality, 18f, 22ff, 51,

Prefix(es), 68-70, 223n52, 225n19 178n63, 178n66, 185n13, 188n17,

anterior tense, 161 194n45

completive aspect, 100, 101 Remoteness. See Tense, metrical conjunct, 68, 194n44, 194n46 Reported evidence, 85f disjunct, 68f, 194n44, 194n46 Reversative. See Adverb(s)

future tense, 160 rien, 176n47

necessitative, 161 Romance, 3, 32, 33, 46ff, 57, 96, 102, 110, pretense, 106 (see also Aspect 120ff, 128, 133f, 138, 139, 142-152,

pretensive) 186n6, 212n77, 224n8, 225n25

progressive aspect, 70, 197n62

prospective aspect, 209n63 sans doute, 12, 18f, 177n53 proximative aspect, 160 Scope, 81, 83, 90, 92, 93, 107, 108, 121—

repetitive aspect, 161 124, 134-136, 204n36, 205n41, Present, $8, 229n7/ 207052, 221n45, 222n46, 222n48 subordinate, 203n30 Scrambling, 108, 117, 214n7, 218n24,

presto, 97, 103f, 208n57, 211n72, 214n7, 219n25

228n17 Semantics, 134-136, 224n17

Probabilitative mode, 54 Semeliterative See Adverb

probabilmente, 12f, 88, 123f, 180n80, sempre, Of, 9, 96, 120, 169n12, 172n26,

180n81, 185n13, 221n43 207n53, 208n54, 221n44

274 SUBJECT INDEX

Sequence of tense, 199n12 ‘constantly’, 105, 222n51 Shortest Movement, 18, 194n45 continuative aspect, 155, 161

should, 86, 198n3 continuous ‘tense’, 96

si, 126, 171n20, 217n15 debitive modal, 158

Small Clause, 168n10, 223n5 derivational, 70, 186n2

solitamente, 4 desiderative, 230n20

solo, 169n12, 180n79, 180n80 distantive (see andative)

soon, 96-98, 135f, 215n10 dubitative, 88, 230n20 souvent, 179n69, 205n39 durative aspect, 158, 161

Specificity, 217n15 ‘early’, 212n74 Specifier(s) epistemic modal, 53, 71, 86f, 163,

single, 44, 184n8, 186n7

multiple, 47, 50, 109, 183n2, 184n8 evidential mood, 53, 71, 72, 85f, 154,

Spec VP, 111f, 114, 173n30, 216n12 155, 158, 174n37, 188n13, 201n22 Spell-out, 111, 124, 127, 137, 215n9, evaluative mood, 53, 71, 85f, 174n37

222n48 ‘fast’, 212n74

spesse volte, 205n39 ‘finally’ 105, 163 spesso, 32, 92f, 173n32, 179n71, 180n84, frequentative aspect, 155, 162, 205n38 204n36, 205n39, 205n40, 205n42, future tense, 154, 155, 157, 158, 161,

221n44 162, 163, 165, 198n2

States, 209n60 habitual aspect, 67, 154f, 158, 161, 162,

Stative,55f 223n4 204n37 Status, honorific, 53f,

still, 95, 174n32, 207n50, 207n51 illocutionary mood, 200n19

subito, 97f, 100 inceptive aspect, 105, 158, 163 Subject, 110-115, 216n12 inflectional, 52, 57f, 186n2

contrasted definite, 112, 216n12 interrogative mood, 154, 162,

definite, 112 230n20

indefinite, 112, 216n12 irrealis mood, 16}

nonspecific indefinite, 112 locative case, 138 quantified, 222n48 iterative aspect, 160, 205n38 (see also

Subordinator, 186n3 frequentative aspect)

suddenly, 208n58, 212n76 modal, 54, 154, 158, 161, 186n4, 186n5,

Suffix 186n7, 187n7, 187n8, 192n34,

abilitative, 54 ‘nearly’, 210n64 (see also ‘almost’ ) agreement, 194n48 necessitative, 81, 161 ‘almost’, 210n64 negative, 222n51 ‘already’, 206n46 non-closing, 52, 53-57 ‘always’, 96 ‘not yet’, 206n46 andative, 105 ‘of uncertain possibility’, 196n56 anterior tense, 67, 157, 162, 164 past tense, 154, 155, 156, 157, 158, 161,

aorist, 198n2 162, 163, 165

aspect, 104, 194n48 perfect aspect, 55, 67, 158, 165, 187n10, agglutinating, 52, 53-57, 186n2, 194n43 190n27 celerative aspect, 104, 212n74, 212n75 polarity, 104

closing, 52, 57f, 185n1, 194n43 politeness, 54, completive aspect, 55, 100f, 104, potential, 81, 155

210n67, 212n75 present tense, 154, 155, 158, 161, 162, conative aspect, 161, 212n77 163, 165

SUBJECT INDEX 27/5

pretense, 106 tout, 7f, 118f, 170n18, 170n19, 176n48,

priorative (see sequential) 219n30

probable, 157, 162, 186n7, 187n7 Transitive Expletive Constructions, 216n12 progressive aspect, 67, 104, 155, 156, trop, 171n19 157, 158, 161, 163, 165, 186n5 tutto, 7f, 10, 46, 100f, 102, 104, 170n15,

prospective aspect, 197n62 170n16, 170n17, 178n63, 178n64,

‘quickly’, 154, 212n74 210n66, 211n71

repetitive aspect, 104, 163, 212n75 twice, 25ff, 179n72, 179n73, 179n74,

resultative aspect, 154 179n75 sequential, 106, 212n78

speech act mood, 71, 85, 161, 163, Unaccusative, 19,

186n3, 186n7, 187n8 Unmarked, 82, 85, 95, 223n4. See also ‘still , 207n50 Default ‘suddenly’ -aspect, 208n58 up, 100

tense, 154, 155, 156, 157, 158, 161, 162, Usitative, 105, 204n37

163, 194n48, 196n56, 230n20 usualmente, 8 terminative aspect, 157, 161

unrealized aspect, 197n62 venir de, 96, 208n56 voice, 104, 154-158 V(erb)-movement, 6, 45—51, 109, 214n5

‘well done’, 101f Verb Second (V/2), 18, 112, 169n12, 180n82, 218n19, 227n7

Telic, 100, 209n60 vermutlich, 12 Tense, 81-83 Voice, 76, 106, 197n63, 212n77 absolute, 79, 199n11 clitic, 197n63

continuous, 96 middle, 19, 90, 102

metrical, 87, 201n28, 208n55 passive, 53, 102, 129f, 197n63

‘no longer’, 206n48, 206n50 See also Passive

‘not yet’, 206n46, 206n48 Volition, 79-81, 90, 203n33, 223n3

relative, 199n11 Volitionality, 203n34

‘still’ , 206n46, 206n48, 206n50 VSO languages, 222n50

Termination, 187n9

to, 189n22 waarschijnlijk, 12, 18, 88 TopicP, 225n25 will, 79, 83 tot, 146, 226n1 would, 190n27 Topicalized phrases, 84, 221n45 wahrscheinlich, 17750 toujours, 6f, 169n13, 169n14, 171n19,

172n25, 176n48, 176n49, 219n23 X-bar theory, 44, 45, 86, 183n2 tous, 116-120, 176n48, 218n22, 218n23,

218n24, 219n30 yet, not, 34, 171n22, 174n32, 206n46